April 23, 2020 § 1 Comment
Gartner has included the concept of Internet of Behavior (IoB) in their Top 20 Strategic Predictions for 2020 and beyond (10/2019). They explain the reasons to this choice:
“The Internet of Behavior (IoB) will be used to link a person digitally to their actions.”
“IoB will also be used to encourage or discourage a particular set of behaviors”
In other words, IoB builds a digital connection to the actions of people, which allows accurate targeting and offering information and services to guide their behaviors. The relevance of this to epidemiological considerations is self-evident and here I shortly consider some of the potentials of IoB for fighting the virus.
When the aim is global and secure monitoring of behaviors, something like IoB is needed to make data collection and use compatible everywhere. Current tools and apps make up a digital Babel. I want to make clear here that this must and can be accomplished in IoB without revealing the identities of behaving people. From the perspective of an ideal IoB, a person taking the vaccination against COVID-19 (in future) is no different from another person with the same behavior (vaccination) in another part of the world. They only share the behaviors and it is a matter of interest how this information is then used and combined with other behavior information, and what these people allow, for medical follow-up purposes, for example.
To put it simply, detection of a behavior, can be accomplished either automatically by any of the current and near future personal gadgets and smart sensors or by allowing people express their behaviors, mental or physical and of any complexity. Clearly, the latter has a major potential for individuals, communities and service providers.
Why would people use IoB? It is meant to secure timely, relevant and accurate communication, offerings and services, better than any AI/ML based system can achieve when it comes to personal and situational needs. In case of a pandemium or other global, national or local threats such a situational intelligence and 100% relevance in communication matter.
I’m not an epidemiologist, far from that, so this is a look of a behavioral scientist.
Coincidentally, the original sources to IoB are my two blogs from 2012 and recently I included IoB in my book “On the Edge of Human technology – An Essay”. I had a vision we could target any ongoing, intended, imagined or planned behavior on earth (like we do with IoT, targeting systems, gadgets and devices) and approach people exactly at the right time and with relevant information and services, when certain behavior occurs – and to do this without necessarily knowing who these people are. The underlying idea was and is that often it is sufficient and beneficial to know the occurring behavior, not the identity of the behaving person. Being a psychologist, it is no surprise I used the form behaviors instead of behavior, which Gartner used.
Fighting the CORONA-19 virus, societies try to follow and predict individual and social behaviors and target citizens with relevant information, instructions and even orders.
Lockdown of behaviors
The global lockdown is meant to control human and especially social behaviors. Because there is no means to know exactly where and how these behaviors occur and will occur, the responsible organizations and specialists provide general, non-specific comments, instructions, guidance and orders. Their hope is that people will follow them. The challenge is to reach people at the right time and with the right kind of information and services.
When epidemiologists offer their specialist knowledge and politicians and journalists add their interpretations to it, media becomes crowded with stories, often conflicting, on behaviors to avoid and instructions to follow: no partying at sport bars, no participating at weddings or funerals, cancelled mass gatherings, use/no use of masks, or no crowding shops. Fresh news emerge about “super-spreaders” who have been known to e.g. share a room with others, going to a certain restaurant, to a wedding, or partying. Large-scale and accurate, real-time maps of occurring or emerging human behaviors do not exist.
The list of forbidden and restricted human behaviors is long but as of yet, there is no exact method to monitor them automatically. The uses of AI and face recognition for example, are too crude and difficult to adapt quickly to the new situation. It would be beneficial if we could get the behavior data directly from individuals, which would allow following them and targeting their virus-related behaviors early, with guidance, information and services to support the virus fight. I want to be clear here: by “following” I mean following behaviors, not individuals – unless they so want or allow it to happen.
It is not known what exactly are the most dangerous specific behaviors: they can be, for example, personal acts or styles, any forms of social and bodily interaction in sports and entertainment, or it can be about the combination of these with physical spaces and their conditions, like ventilation, hygiene facilities and practices where close contacts happen. Unfortunately, tracing such behaviors post-hoc or tracking them in real time is very difficult or impossible. For example, why isn’t there a service in restrooms that when a person tries to leave it without washing hands, it would have a way to remind of it? This would require behavior coding like in IoB: someone can visit a restroom and use the mirror only J
Schools have been closed in many countries, but now they are slowly opening and a new knowledge need emerges there, too: students and teachers are instructed to behave in certain cautious ways but it is difficult to collect data on what actually happens and what consequences different behaviors have. Specific tools are needed and IoB or similar solutions could help in this.
A flood of news mentions certain individuals and their behaviors (super-spreading) that boost the spread of the virus. For example, Dr. Hendrick Streeck told The Guardian that where there is dancing and singing, the virus spreads fast. However, the burning question remains “What exactly are the virus spreading behaviors”. There is more to our behaviors than dance and singing.
Better and informative behavior data is accumulating fast and under global scrutiny. When the most ‘dangerous’ behaviors are known, people can be targeted and guided accordingly. In many, if not in most cases, this could be done, based on their behaviors, but without knowing who these people are.
As expected, apps with standard technologies like GPS and Bluetooth have been quickly introduced to monitor the movements and whereabouts of infected or potentially infected people. Knowing where they move or when they are in the vicinity of others is hoped to help prevent contamination and spreading of the virus. Anonymity and voluntary use of the apps and tools are emphasized. People have had the bitter lessons on how their personal data is used by network giants and their networked partners and are getting now eager to protect their privacy even and especially during a crisis like this. Indeed, there are reasonable worries concerning such detailed identity data if it can be used for other purposes, for example, by insurance companies, financial services, or by any other sectors benefitting from intruding people’s private life.
Some technologists doubt the use of their location (GPS) data for tracing purposes while others have seen problems in interpreting Bluetooth –based data. No doubt, these problems will be solved, and the tools will be useful during the post-virus recovery and follow-up time and then of course, for the next crisis to come. https://www.bbc.com/news/technology-52353720.
What could be relevant behavior knowledge?
Behavior knowledge is more than what we directly observe of a person. For a psychologist, it is natural to include planning, emotional experiences, interpretations and intentions, for example, under the term ‘behavior’. IoB is meant to cover and ‘code’ any of these human phenomena. Much, if not most of our behavior is internal and relative-to-others in nature. We can know the exact location of a person, her movement patterns, or participation in gatherings but still fail to understand her or know the intentions and motivations behind the observed behavior pattern. From the outset, people can behave in similar-looking ways but for completely different reasons.
Present digital technologies monitor simple observable behaviors, which has its undeniable value in entertainment, sports, health and life-coaching apps. Interestingly, Spotify can be considered as a primitive form of IoB since it lets people express their wishes and to be rewarded by relevant music when they so want – the behavioral loop is inbuilt in it.. Much more could be accomplished with IoB there as well.
With mature IoB it becomes possible to detect any behavior, external and internal alike, that can be coded or expressed so that when it occurs, the person – her media environment – can be accessed accordingly. As human beings, we are the best experts of our minds, of our experiences, intentions, value priorities, and mental states, better than any AI/ML system. Simple and effective means to indicate this mental information has huge human and social potential and invites to accessing each other at the right time and with relevant messaging and services. By ‘accessing’ I mean the possibility to approach a person via any of her own app or tool or by directing information to her visual/audio environment.
An exceptional form of knowledge in preventing the spread of the virus would be reliable intention knowledge or knowledge of a person preparing to do something that increases the risk of infecting someone or to be infected. In web and mobile apps this can by arranged by asking the person using the app but this is a cumbersome maneuver and does not ‘live’ with acute life situations. More dynamic apps are needed to make IoB ‘conversation’ seamless and even fun.
Many would argue that people are unreliable in expressing their intention data. However there are masses of human intentions that predict certain behaviors with a very high probability, and it would be useful to have this knowledge. It’s no rocket science that a person knows when he’s going to eat something, take a medicine, visit a friend, before he is there, sometimes days, hour, minutes before it. From the virus perspective, it would be extremely valuable to have this pre-knowledge and to inform him or other people from that future or near-future behavior. In case of hospitals and health care services in general, there is much more to it – this knowledge can be critical if we had it. The benefits scale up with masses. How to get access to such data and would it be safe?
Why is behavior data linked with the identity?
Prominent public figures have expressed their fears that the pressure to track people can become a permanent social practice with dire social and political consequences. Their worry concerns technological solutions, where identity data is intimately linked with other data of the individual, as is the general practice today. However, this need not be so.
Most digital apps and tools that use our behavior data, like when messaging, traveling, visiting places, using services and exercising, indeed record both the person’s behavior and her/his identity and uses this for various purposes. Historically it was a dominant social practice in public, financial and health care services to trust our identity data to these services. This is how behavior data became de facto behavior+identity data and most analysis tools use both of these components to target the campaigns, services and any lures of life.
Interestingly, still some 50 years ago, there was not much public worry about id data, but then the net changed everything. Anonymity is now added in case the users especially want it, but it is far from a standard feature of the services we use and most of us don’t have a faintest idea where our identity data is stored, shared and how it is directly or indirectly used. GDPR helps but it is only a superficial cure to this paradigmatic trend.
Getting relevant behavior data – with IoB
Originally, when developing the IoB concept, I imagined the following:
“What if we could know, when, at a specific moment of time, certain human behavior occurs somewhere on the globe, and we could be in touch with all these people, from single individuals to millions, behaving like that, but without knowing who they are or where they are unless they voluntarily disclose their private data?”
How could we get that data and what could we do with this huge global and local pool of behavior knowledge? Internet of Behaviors is meant to support services, which systematically record, code and use behavior data, which is not automatically connected with the id of the behaving person.
In the case of COVID-19, the behavior data collected via an IoB app would allow monitoring single individual or communal behaviors occurring right now in masses. This can then be supplemented with relevant context data like geographic, organizational, process, community, medical, economic, or any other background information that allows mapping the ongoing behaviors on whatever is the context or domain of the behavior.
As a simple acute example, a pharmacy could offer its own IoB app to its customers and let them indicate (web, mp) when they plan to visit or are on their way to the store, so that the store can be prepared, especially if crowding is expected or a Corona patient’s family member is coming to fetch the medicine. There are various ways this information can be used at the pharmacy but also within a broader local context – without revealing the identity of the ‘behaving persons’. Communication is two-way which allows the pharmacy to send information to the person(s) but again without necessarily knowing who he/she is unless they have made an arrangement in the IoB to share private information. Then of course, the IoB approach has a multitude of uses elsewhere and scalable.
IoB can provide predictive information – about any intentions and plans, related to the situation at hand, like COVID-related behaviors. When a large enough pool or crowd of people use an IoB app, it becomes a tool, for accurate forecasting, This makes it different from any app aiming at following movements, tracing people, detecting locations and occurring closeness of people. Together, however, these approaches can make up a very powerful and situationally intelligent service. IoB can be integrated in these recently published tracing apps.
In practice, with the IoB app, the user could easily indicate or select from a set of alternative behavior codes or have a link to any personal QS gadget like Oura ring, or use QR devices, or otherwise indicate the ongoing or her/his intended behavior and so on. It is a matter of innovation to figure out the most feasible, seamless, automatic or semi-automatic ways to let people indicate their behavior in IoB, with a push of a button or why not with spoken comments.
IoB has exceptional power if it is developed in coordinated and standardized manner so that it makes coded behavior data accessible globally and makes it usable for any possible purposes. Apps can be designed for limited use like the pharmacy case above, but already there, the data structures should be designed so that they allow significant scaling up. Originally I imaging that IoB could use IPv6, with a pool of dynamic addresses reserved for certain behavior classes and which various service providers could then use and distribute. This is not mandatory and other forms of behavior coding is possible, for example so that it makes possible a private, even dynamic definition of behavior codes between the client and the IoB service provider.
Some aspects of design
Basic IoB features can be easily integrated or built into current web and mobile services and tools. When anonymity is required, it can be added. However, IoB could be built on its own architecture, supporting environment and protocols, just like IoT. I am tempted to imagine that because of its versatility and the way it can touch practically all aspects of human and social life, an IoB device could find its own independent place as a personal IoB-gadget – not part of a smart phone – or as an integrated part of any smart devices or ubiquitous ict. These visions are exciting.
I have given a superficial description of the IoB as I see only some of its potential value in fighting a global pandemia or other global threat . This is by no means a product or service description, but I hope it inspires design thinking for developing IoB in practice.
Gartner (2019) Top 20 Strategic Predictions for 2020 and Beyond.
Nyman, G. (2012) Internet of Behaviors.
Nyman, G. (2012) The Psychology behind the Internet of Behaviors.
Nyman, G. (2020) Behavior data in the net. In: “On the Edge of Human technology – An Essay” Amazon.
April 17, 2020 § Leave a comment
Every collaboration session over the net makes up a drama. Surprising enough, most network interaction and collaboration tools and services – visual, audio, text, graphic, data and communication structures – in global use do not amend from the knowledge and experiences from movie and theatre directors to secure seamless interaction sessions. The world is full of collaboration directors. This is surprising: as movie watchers we are utterly sensitive to the story and the drama, its characters and people, the style, depicted environments, camera work and the sound-scape with music.
Many of us watch movies and often recognize the director’s fingerprints, especially if the movie does not ‘work’. We are masters in judging that a movie is boring or captivating, if it feels unnatural and the characters are not convincing; we recognize the lame stereotypes and disturbing cuts or if the story development is poor or unsuspenseful. Sometimes the first few minutes of a movie are enough for us to judge the quality of what we expect to happen. On the other hand, we have favorite directors and we are every now and then positively surprised when we see an exciting story and style, a hint of a novel genre, great camera work, satisfying directing and acting in a new movie. There is no reason to assume that we are not susceptible or sensitive to similar decisive experiences and judgments when we join a collaboration session over the net.
I’m not a movie director although I have some background in theatre lighting, which was a 10-year lesson on what makes a play or a movie happen on the stage, on a screen and in our minds. Only a few days ago, Lauri Törhönen, a prominent Finnish movie director reminded me of some disturbing side effects that come from haphazard use of present communication technologies: static scenes, stiff postures of people, bad lights, poor use of camera and shooting angles, movement noise, bad audio surrounds and so on, which together generate “a bad play of their own”. In such cases there is no solid support to the story, the people, to their personalities, roles, styles, and motivations in what they do in the collaboration sessions. The message becomes unavoidably twisted or biased by this unintended drama running free. These are simple and superficial aspects of ‘collaboration directing’: there is, of course, much more that we can learn to build a story and a drama we want to happen and to pay respect to the story we want to tell and hope our partners or clients receive it with interest and motivation. We can take lessons from competent and experienced directors.
We don’t know the actors but they still touch us
Most of us have a relatively long history, more than one year, with the colleagues, partners or clients we now meet remotely. This can make communication easy even with primitive tools. Our shared history has a built-in, predictable psychology of its own – a psychological hysteresis – something we rely on, enjoy or suffer from. In everyday life, we don’t have to be consciously aware of this backbone of our shared behaviors, expectations, and attitudes.
The power of ‘relationship psychology’ is exactly this: we don’t need to put too much effort on checking the status or state of our relationship with colleagues or long-time customers, every time we start communicating. Only in exceptional situations likes crisis, in matters of opinion, or personal disagreements we must somehow consider this more deeply, when we try to find out new ways to behave or are puzzled by our own reactions to something in the interaction and the relationship. With new acquaintances this is more complicated and especially now in remote interaction we must try to ‘sense’ or know, and build the relationship.
At work, the psychology behind our behavior emerges around the drama and story concerning the task, the environment, and the people we work with. For example, you can be a member of a sporty sales team with its own work history, private behavior style and language and a set of (psychological) sales tools and concepts used when meeting customers. The sales situation, like any other work situation, from the beginning to the end can be considered as a drama consisting of familiar or routine episodes and it is played with specific roles.
Sales persons and managers alike have their roles in the sales drama and can spontaneously participate and behave according to it when meeting colleagues or customers f-f. In remote work the drama happens in different, distributed settings. The sense of presence of other people, ‘the stage’, the impact of roles, and the way the play can be guided and proceed, have all been transformed and constrained by the technology used in communication and as a result of the practical arrangements of the virtual sessions. Like in the case of movies, the technology alone does not make the play happen: the first classic b & w movies, with their primitive technology and caricature form demonstrate the power of the drama and the story and we can still enjoy fully the primitive movie form.
No need to emphasize this: building and directing a compelling story and the drama, is of utmost importance to successful interaction. When this is not deliberately done, a drama of its own emerges anyway and it may not always be what would have been hoped for.
How would directors advice us in the use of present technological means to build a compelling remote drama of collaboration and work? Could we learn to ‘direct’ the remote sessions so that they can better serve their core purpose now and in the future? Can we design better social-technological solutions by taking seriously the knowledge and experiences from directors? Have we taken our lessons now and start designing a new genre of tools with human touch, post-virus?
Here are my first guesses on the directors’ advice, based on the inspiration from the sites of Raindance and IndieWire, but you can easily find dozens of similar suggestions.
- Remote work sessions and the whole process of interaction are built around ‘technologically possible and programmable’ human behaviors and experiences, many of them familiar from the f-f- world but some new ones are introduced in the remote situations having now their own ‘life’ an indeed, often it means ‘family life’.
As an example, the family life of our colleagues, clients, teachers, and partners has never before been so intimate aspect of our interaction sessions and with this scale. Sometimes the impact of this can be seen, heard, or sensed during the interaction and even before it perhaps, but the psychological presence of the families, homes and their spirit is practically unavoidable. The members of a sporty sales team might reconsider their style of talking when there is a family member listening behind the back while the same can be happening at the remotely connected home site of the client. Ways to deal with privacy and trust must be openly and specifically declared and to make it compatible with the psychology of the remote work drama. In the present work and education context, privacy and trust have their new or transformed psychologies, not totally unknown before by certainly different now.
- Every episode, even the smallest one of them, is or becomes part of the story we tell and present. It is up to the manager of the collaboration or of other form of interaction session to decide how each remote session episode should contribute to the story. To become aware of these episodic effects it is important to know the participants, the gist and the flow of the story and its core episodes. Roles, styles, progress and the outcome must be guided, and not to forget the end – and “what next”? Before this can be done, all these elements must be defined and described, in more or less detail.
- Build a drama you want, for every session. Even a rational, collaborative process with a straightforward agenda lives as a drama we experience and it has psychological after-effecs, positive, negative or neutral. It can have any form, from extremely competitive and witty play to a wonderful, almost romantic demonstration of mutual love and respect – in business and in any other activity. It is up to us to guard and guide this, or whoever has the power and chance to make it happen.
- Make it a systematic task to follow the emergence of the drama of each remote session. This helps in recognizing what actually happened during the drama and its episodes. Elements of success and failure can be seen as well as early warning signs of problems.
- Construct a model of the way you make your team/firm/community act in the story and the drama. For a challenging, large-scale meeting or collaboration, a storyboard helps to recognize its components and analyze its functioning. Don’t do it as a bureaucratic exercise, but a form of team learning. Take time to prepare for each session or episode; it need not be hours. A reasonable time invested in it pays back.
- Help your staff to learn about themselves as acting participants in the collaboration play. We are not natural actors and a professional director or other relevant specialist can make this learning progressive and fun, even and especially when arranged in the for of a real remote session.
- Don’t forget the feel, style and spirit of the remote working drama. Often it does not happen automatically, but it can be made to happen but it requires roles from the participating people, jus like it happens at f-f situations. Music is not forbidden during the sessions and why not use it as a form of relaxation during the pauses or to signal certain events, their style or the phases of work. However, must be used with care.
- Unlike in plays and movies, there are characters that are reluctant to participate in the drama. They can be your customers, your business partners from other firms and cultures, or your colleagues who are not comfortable with such a ‘play’ or ‘role games’. Make sure they too can find their natural behavior styles, don’t use the tern ‘role’ then.
- Cultural differences matter. Take the word ’culture’ seriously, in its widest and deepest sense, from global to local aspects of human life. We have our preferences on what plays and movies we enjoy to watch.
- Finally, the prelude. When we go to opera, the prelude invites us to the spirit and atmosphere, even the drama awaiting us when the curtain opens. In most forms of music, it is an essential part and we have grown experts in recognizing the pieces from the first few notes and chords. There are no general practices to do this for remote sessions at work and in collaboration. Hence, remembering the impact of the prelude, this is an invitation for creative solutions to this, how the prepare for a remote session, and to do it properly, according to the style and spirit of the firm or whoever engage in remote interaction. It can be quick moment, lasting for a minute or less but it can be hours, even days when it has a solid purpose.
A call to network operators, service providers and app designers
From a purely technological viewpoint, transforming the families and family clusters into the nodes of local and global networks may not look much different from adding any other clusters to a working community and start using the net and mobile services. However, families living now in quarantine amplify and reorganize the human and social aspects of all network activities: home and social context becomes an inseparable part of all organizational and any personal network activities. Families are units that make choices and decisions, they maintain certain ways of behaving and they support and motivate each member. Firms and organizations must consider this as a fact of their new life and it has consequences.
Taking the family-related network transformation seriously, network operators can find guidelines for developing their tools and services and make it truly human and social technology. The present transformation should open the business-interested eyes to see the networks of life on several human and social levels, with all the human possibilities this offers, and to remember the directors’ lessons. Then there are challenges:
Firstly, current network tools and models have problems in serving user clusters working simultaneously with very different contents and in variable (dynamic) situations and contexts. Most of the tools have poor if nonexistent human-situational (contextual) intelligence. For example, parents working remotely for different firms must do it in isolated collaboration systems or ‘virtual boxes’ and tools that barely communicate with each other. They exclude any seamless and human interaction with children. There are several security, brand, cultural-historical and technological reasons that have led to dominating use paradigms, which have not been massively challenged before now.
There are no simple technological solutions to serve people when they change their ‘personal space of interest’, at work or in pleasure. The same is true with the isolated digital tools that children use in their studies and parents use for communicating with teachers and school administration. In our everyday life, we just move our focus of attention – for example, from the ongoing task to our children – and can do it in an enjoyable and spontaneous manner: “John, why don’t you come here and tell me what you’ve been doing?” With present tools this is not so. Now, during the quarantine the digital Babel of tools is revealed when all these systems must be used simultaneously, with practically no coordination or communication between them.
Secondly, many communication difficulties originate from the poor or nonexistent situational knowledge the communicating parties have on each other. This causes bad timing and targeting of messaging, lots of extra explanation on what is the situation and its demands, and then of course, irritation from these uncertainties that disturb fluent work, interaction, awareness building and prevents reaching people at the right time with relevant information and actions. We have practically no way to know the psychological state of our communicating partner, just before the remote session. The prelude could find a positive role in this.
Thirdly, instead of one or even several networks where people work at a time or in parallel, every family is a conglomeration of several, intertwined and overlapping networks that cannot be separated from each other without distorting some important aspects of their life. For example, children who now study and do their homework are not psychologically or even physically separated from their working parents. Continuous physical interaction becomes problematic and both children and parents can suffer from this. New practices must be learned and it takes time. Digital and network tools don’t support this – although they could include many family-life orientated services and features, which understand the drama of everyday life. The dominating digital paradigms do not offer tools for people ‘living’ in one network to express their ongoing (behavioral) situation to people in another environment and network. Such a situational knowledge, however, is critical for seamless communication and interaction. Messaging, document and image sending or sharing is of course easy and even seamless, but it is only surface.
The risk for family problems, even in families who have not suffered from them before, comes from the need of parents and couples to manage their jobs and responsibilities and to be tied to their digital tools and their demands that link them to their work. The tools are not designed to support our best human qualities in families and other intimate contexts. The best of the video conferencing services are rational in nature and support ‘lean’ collaboration and interaction. They are not designed for intimate and deeply personal matters of life; other tools are available for that. To exaggerate a bit, we use rational tools of a strong technological paradigm but now we can see the real value of the ‘soft’ elements in human communication. Listening to the families of this new global network can offer striking possibilities for better human-technological advances.
In organization life, these human aspects are no less important and it is only due to the digital paradigm that our tools are ‘cold’, insensitive or harmful. Many think that people have bad ways to use social media, which causes the familiar problems from bullying to aggressions. Another way to look at this is to see it as a human-technological design that helps and invites such things to happen, without constraints or control. The bad human design causes them, it is not only allowing them to happen. The globally emerging family nodes have a human message to tell to the networked world.
Let’s not forget the leaders and managers. We know they are under pressure now, but they need social and technological support, too. I’ll come back to this later.
April 6, 2020 § 2 Comments
Moving to full remote or virtual work, and re-organizing network activities does not happen without human (psychological, social, health) costs. In addition to the firms paying for these costs, a heavy burden comes to the individuals, families, and even their relatives who help in keeping businesses and service alive. I will not deal with the destiny of the unemployed here; their suffering is different now and firms and organizations should not forget them either. It is a story of its own, how to do it and why.
This is important for managers and personnel to understand and build awareness of it: we face a months-long period to learn new habits and practices and to get rid of the harmful ones in how we communicate and collaborate within the net. It is not only a human challenge – there is the imperative to maintain sustainable business. Hence, here I introduce a sample of the many problems that can be easily forgotten and become serious hindrances to work, but which can be overcome with wise actions. Every firm must find its own way to manage the new (virtual/remote) situation. It is not wise to adopt any practices of network life without serious consideration of the specific context of the firm, and especially its recent history, culture and values. Perhaps the most important advice to the management and personnel alike is the following:
Do everything you can to preserve the best you have in your firm and your people, in the way you have lived and worked together, served the customers and committed to the common cause.
Taking the above as a base for development and care, there are simple and practical ‘human-social factors’ at work to be considered:
- Cognitive load increases significantly in remote work, for many reasons, for example because of the extra, continuous attentive burden and complex messaging practices. For example, in normal work, we are aware of the social situation and people around us, and have simple ways to take a mini-rest and well-timed privacy, have various forms of support and enjoy the relaxing predictability of communication and interaction. We can use our limited attentive resources wisely.Long-lasting remote mode makes a fragmented work environment and becomes a new burden for the personnel – there is the continuous need to follow, expect, to monitor, to be ready to react to any coming messaging. Our memory in all of its functions, from the short-term to long-term processes, to memorizing tasks and events, episodes and meanings, becomes crowded. People get tired of this multi-dimensional psychological pressure and it has consequences that are not always immediately visible. It is a good practice to take time after work to consider this and be sensitive to any signs of stress, fatigue and other psychological or physical symptoms. It is wise to share these personal experiences among the personnel and friends – proactively – so that we learn about them as a community. Discussion with colleagues is helpful and we can learn from their ways to view or ‘frame’ the situation and to cope with it. Technological, organizational, working model -related, scheduling and other actions can then be taken to relive the situation.It is not easy to see the direct causes of cognitive burden, because often the symptoms occur as a fuzzy ‘internal itch’ difficult to perceive accurately and to manage it. Sometimes these problems become tangible when several tasks must be accomplished at the same time, which requires all the available mental resources. However, it is possible to become sensitive to them and try to organize work, practices, scheduling, private life, and rest accordingly. Proactive preparation has extra value and helps to be prepared. Working at home introduces new psychological, continuously present phenomena to family matters: child behavior, all routines, interaction and many others demand their deserved attention, care and – again – our limited mental resources.
- Extra cognitive and attentive overhead of textual communication is an essential aspect of almost any organizational communication of today, either before, after or during the communication. This is caused by the simple act of writing which takes time and effort to do it properly: to correct, re-correct and read requires full attention and takes energy and mistakes and bad expressions can have major negative consequences. People have different writing skills and harmful communication styles and related misunderstandings matter, as we too well know from social media. In this sense, social media is indeed an excellent environment to learn what not to do and how not to communicate. Care is the core, at all levels.
- Isolation from the physical and social context means isolation from relevant and dynamic information and support, much of which is tacit, that is, we have learned to take it for granted and it is almost invisible to our everyday life at work. For example, a colleague in the same room is not only a person sharing the space or process with us; her/his presence means he/she is physically and psychologically available. The tacit knowledge of this is relaxing, we even enjoy it as a friendship or of having a colleague close to us, and it contributes to the feeling of trust, comfort, and support when needed.When in need for help or any other interaction, it is a simple process in f-f to initiate: we have learned to be sensitive to all aspects of such human relationship management and interaction. However, in remote work and virtual communication everything changes and extra checking, messaging, preparation for contacting and initiation of communication is needed. Present collaboration systems lack proper support for this kind of profound, human phenomena and behavior. The psychology of intimate communication changes profoundly and we must remember that.
- New problem behaviors emerge – one among many being passive style of working. A passive person in an f-f team can be easily identified and invited to take action, guided or controlled. In virtual work, this style is not easy to observe and it takes extra time and control efforts to notice it; time and efficiency is wasted and irritation generated because of it. Sometimes people are not aware of their own ways of behaving in the net and they need a ‘social mirror’ to learn. Predictability of behavior in the whole organization is worth gold because it helps people to be prepared, relieves from extra control and it is simply pleasant to work then.
- Management of virtual/distant activities requires skills, knowledge and experience. It is different from standard ways of management. It may be necessary to reconsider some organizational processes, roles and structures if/when remote work continues over long periods of time and in 100% mode. The functions and responsibilities of management teams need reconsideration since virtual communication requires extra effort and time, and a large part of it does not follow standard organizational processes and information flows. New information flows emerge and as a result, some managers are at risk to become overcrowded by communication.Interestingly, already in 1980s we found out in a study that people are extremely bad at estimating the amount of pressure their managers experience in their communication. People overestimated by a factor of 3 to 10 (if I remember correctly, it was a lot) how much information they thought they could personally send to their managers without causing disastrous message crowding. Blindness to system effects was evident and it is a common problem today, too.
- Interaction dynamics among personnel change because of the way remote communication happens in practice. People, who have had a significant positive and valuable contribution to these dynamics at the site, can lose this role and become a significant loss to the firm/team dynamics and atmosphere. This is the time to see the value of these great people who often do not get recognition for this contribution. It is good to look at other important roles in the firm as well and make sure their contribution is not lost. A psychological role inventory is extremely useful – and it is rewarding for people when they can talk about what they really do.
- In customer work, virtual tools can be a problem and even a new risk, not less because of different communication cultures in firms. Extra sensitivity to communication style, timing, and targeting is now required because of the network context – and the fact that customers, too live exceptional times in their own work and family lives of their personnel. The history of trust in these relations is now valuable psychological capital and it must be properly recognized, in all its different forms. A good advice is to offer relevant proactive information to the clients and then invite them to communicate towards you so that they have the initiative.
- Network and collaboration tools have taught us to mix synchronous (phones, audio & video conferences …) activities with asynchronous ones (emails, text messages, podcasts, printed docs…) and often it happens that people are not aware a) of the different time constants and other systemic consequences these tools have (how soon and when a response can be expected, what other communications and actions they generate), and b) of their timing-sensitivity (when is a good/bad time for certain communications) and of various interactions of these. In remote communication and work, the mixing of synch/asynch communication can become a distraction, a source of uncertainty, distress, a disturbance and introduce new negative, psychological forces that disturb seamless interaction.Simple examples: When you prepare emails the previous evening, don’t send them immediately then (and think that they will be nicely in the piles of emails waiting), but time them to the morning or some other relevant time at the office or at homes. Furthermore, your team can decide on certain times of the day (lunch break, for example) when no emails or other messages are sent or virtual chats and meetings are arranged. This synchronizes behaviors and relives people from continuous monitoring of their team-tools. It can silence the traffic for 30-45 minutes, for example, which is not a long time but can be psychologically significant when used accordingly. For people working from homes, the importance of this is magnified.
- Sleep and rest. Tune the remote work arrangements and communication so that you can enjoy proper rest, sleep and regularity of everyday life. Some have already noticed how the quarantine has allowed a surprising rest when 1-2 hours of commuting is excluded and have then observed the quality of work outcome to improve. However, this does not happen automatically and without systematic actions to arrange the work. Make sure that work does not conquer every hour or unpredictable times, and cause harm and introduce conflicts and other nuisances to good family life. Private, team-related and organization-wide schedules, even dynamic ones are need for collaborating people; this is not a matter of individual arrangements only.
The sense of presence
When we, managers and personnel alike, start working remotely from homes, we lose the strong and real sense of social presence of the people we normally work with, more or less intimately. This loss is not insignificant. The feeling or presence is like invisible or tacit ether that carries us in social life. It is where we live with our colleagues and can sense each other’s closeness and availability. We trust that many of the colleagues are ready to offer their immediate help when we need it. We expect and encourage them to approach us and we love to share the joys and challenges of co-work and existence, to help when we can. They know we are for them, when needed.
In remote work, the sense of presence becomes fragile and weak over time unless it is cared for in an effective (and affective) way. Collaboration platforms and teamwork tools vary in how they contribute to and maintain the sense of social presence but in general they massively underestimate this aspect of human life and it colorfulness. There is much to improve in these tools and services; their psychological outcome depends more on human behaviors than on any aspect of the technology itself.
Inspiration from game experiences
There is a special field of technology where the human-social aspects of human experience have matured fast and shown their power: computer games. Many of you who actively play computer games have been surprised by the power of the psychological immersion they induce in us. Now that ‘the virus game’ takes the form of technologically organized remote work, it is good to consider some of the psychological secrets behind the game experiences and see what could be learned from them to improve the remote work between homes, offices and firms.
My colleague Dr. Jari Takatalo, now at the game company Rovio Ltd, prepared and integrated a wonderful collection of known psychological factors into a systematic and holistic framework to describe the psychology of game experience. We (it was Jari’s ‘baby’) introduced the multi-dimensional model called PIFF (Presence, Involvement, Flow -Framework), which consists of the critical experiential-psychological dimensions that make an enjoyable (or terrible) game experience. The model captures the critical psychological experience variables, that is, the quality, intensity, meaning, value, and extensity of an experience and it is intimately grounded in the work of the psychology classics like James and Wundt.
In the following Table, I summarize only some of these variables, the nature of the main psychological dimensions as they occur in PIFF and relate them to the psychology of remote work. The idea is not to copy the model for describing the experiences but instead to remind of these important experiential dimensions. Together they contribute to positive (or negative) psychological outcomes of remote work if they are (not) recognized and arrangements are (not) made to serve them. Many of these psychological phenomena are relatively easy to recognize when time flies and work feels exiting and fun. Remote work has its own, complex contexts and generalization is not wise, but it is good to look at some of these psychological background elements.
Discussing the topic of movie direction with the prominent movie director Lauri Törhönen in Finland, I was inspired to think about ‘directing a remote session’ and will tell the story in the next blog.
April 3, 2020 § Leave a comment
Creative Commons Zero
Corona virus has forced both public and private sector organizations in Europe and US to move to remote work – to start working from homes. Surprising enough, an essential aspect of this abrupt transformation has received little attention among the world leaders and leading media: never before in human history of networking has there been such a massive, fast, and global transformation – or experiment – in the way human work is organized, managed, and made happen at homes and households. (for an early insight, see e.g. Atlantic). This is not only generating new social, work-related and network practices, but it disturbs markets, creates new ones and inspires to new businesses that adapt to the quarantine situation. Some of the new services and businesses will prevail even after the pandemia.
Hundreds of millions of families give new life to networks
In EU there are 220 million households and 130 million in US. Typically, about 20-30% of them have one or more children. About 15% of the EU households are single parent families while in US the proportion is somewhat larger. Assuming that more than 50% of people now work remotely, the number of families living this transformation comes close to 200 million homes in EU and US alone and tens of millions of children and their education become affected.
The transformation concerns practically all firms, public and private alike, even those where working at physical sites cannot be avoided. At the writing of this, EU and US – skipping the other parts of the world – are reorganizing all their networked functions and learning on the way. Questions now arise: what does this mean to work – now and in the near future – to its form, management and content; what happens to family life, its economics, everyday practices and atmosphere, child behavior. What new problems emerge, even in families which have not suffered from them before, like alcoholism and drugs, home violence, anxiety and any harmful dynamics caused by the new pressures.
Bending a bit the famous quote by McLuhan, the homes and families have now become part of the digital messages – not only messaging – and communication. The nodes of the emerging EU & US networks of work are the millions of homes, each one of them connected with one or more firms – and other families, family members, significant friends and partners. Network dynamics and the distributed value base there change profoundly. Nobody knows the exact consequences this has on the way working takes place, how people and their leaders interact, how things are coordinated and how the work output is related to the new situations at homes.
Software development has not been prepared for this. Looking at the top 20 collaboration tools available you can see that they are made to support projects, task lists, conferencing, documentation, messaging, scheduling, while they almost totally lack any features that would be aimed at building human connection, support, positive atmosphere, and care. They have nothing to offer to parenting.
There will be more family negotiations than ever before, touching any aspects of work topics relevant to the parents and other family members. New forms of conflict appear at work and at homes. This is not the place to deal with all the psychological questions and problems, and I offer only some observations and thoughts about what is going to happen next. Lancet (Vol 395, Issue 10227, March 14, 2020) has presented an extensive review on the psychological problems of quarantine: The psychological impact of quarantine and how to reduce it: rapid review of the evidence.
No relevant network model to describe this abrupt development
Network technologies are now under test and they seem to manage the traditional traffic rather well as told by Anne Morris/Editor in Light Reading (23 March, 2020). However, the pandemia puts the life of families and firms in another test during this exceptional period. At the moment, there is no network model available to describe this dynamic and the new net life – its connectivity, layers of activity, behavioral constraints, work content and context, the loves and joys of family life and work, the nature of interaction within this new complex entity. It is difficult to see what happens in this new network and why, although anecdotal observations on the new life at homes crowd the media. Some parents in Finland, for example, have been surprised to note how here, in the land of world-class education, home schooling can make schoolwork more effective.
Think about the human networks of a family: it has layers and clusters for friends and relatives, but it has other layers as well, for professional activities and hobbies and then of course, the layers of the job-related tasks and information sources and the firm and so on; it is a multi-layer, local-global, interacting network system, a very complex creature indeed. Now that work is conducted at homes these layers and clusters become increasingly intertwined, and the dynamics of their interconnections are not easy to model and describe. Hence, it is unclear how to model and measure their functionality. Traditional traffic measures can be easily applied, but they miss the human point.
Padgett (2007) applied a multiple network description for modeling organizational genesis in the historical Florence. However, the scale and extension of the family-related networks currently activated, and the technology used are significantly more complex than those in Florence then. There is no methodology by which it would be possible to ‘introduce’ relevant content, human-social meanings ad interdependencies into the network of such human and technological complexity. The contents and meanings are different for the firm and the family, for school children and their parents and for the communities of their friends, for the colleagues and management at firms and other organizations. It is almost an impossible phenomenon to cover with any models. We need relevant “toy models” for research, and practically a new research field and paradigm.
This may be a surprise: there is no general network model that can represent all these layers and contents of human activity that occur simultaneously, interacting. With my colleagues we have earlier looked at a related problem from the perspective of the networked firm and evaluated the feasibility of alternative network models, from simple straightforward nets to multi-layer and game networks to describe the life of a networked firm. None of the models we considered seemed feasible to model a simple networked firm properly. On possibility we saw was to use a locally sensitive, multi-layer model, combined with game network dynamics to cover the significant psychological aspects of the network reality, the drama, in firms. (In: Big Data and Smart Service Systems, “On the Behavioral Theory of the Networked Firm, by Nyman et al., 2017). However, what happens now in the global network of families is different. In my recent book “On the edge of human technology” I have speculated about the use of a totally new type of network description suitable for this kind of complex problems, using an analogy from string theory in physics, but so far it is pure, very pure speculation.
Only two months ago, when working at a physical site, the organizational and management model, company culture and its value base guided and constrained us in our work behavior. In the present situation, families and their ramifications introduce new forces into the work context, directly or indirectly, which introduces new values, new constraints and new content into work. It may not be immediately visible at the organizational level, but at homes and in families the situation is an immediate, everyday fact of family life that must be dealt with, to reorganize the way things are run at homes. As an example, millions of single parent families having small children adapt to this, without compromising the well-being of their children. There is no standard way to do this and national and family cultures differ significantly between nations in EU and between EU and US. There is no doubt that these new inputs to work have both positive and negative potential to contribute.
Then there are the managers who must cope with this new situation. Let’s not forget them.
February 27, 2020 § Leave a comment
(Image from Piqsels, License to use Creative Commons Zero)
Göte Nyman, University of Helsinki, Finland
Ossi Kuittinen, SimAnalytics Finland, Ltd
Deep learning AI systems have, until now, been black boxes and although they show amazing high-level artificial intelligence, it has become imperative to understand their detailed, underlying logic, the computational elements in their critical decision-making, and their functional structure in order to know how and why they end up making specific decisions. The same need for transparency concerns the learning phase and the choice of the materials used for their teaching. The problem is not new as it was met and recognised already with the logical and symbolic AI developed by the pioneers like Newell and Simon in 1950-60s.
The reasons for this emerging need for explanatory AI (xAI) are evident: the costs of failures and underperformance become intolerable when AI enters large-scale and sensitive domains like medicine, traffic, weapon industry, and massive industrial settings. Explanatory knowledge can be critical from various perspectives: what limitations are introduced by choosing specific teaching materials, how the AI system generalises from this, what happens in exceptional situations, what kind of failures are possible, how to organise communication within the AI system at large, are there ‘black holes’ in its learning, and how and why should its parameters and other performance-related variables be adjusted for a better, optimal or safe performance and finally, how should xAI communicate with the humans and the human community or teams working with it.
There has been the underlying assumption that accuracy of computations in large AI systems prevents explainability but this is now being challenged. Accurate, interpretative deep learning (DL) models are being developed and tested where their reasoning is modelled e.g. by imitating the decision making of professionals in the specific classification task (take a look at e.g. image recognition task in Chen et al., 2019). However, the problem is still open and new developments can be expected to occur soon.
In psychological and social sciences, there is a well-known tradition in modelling human decision making and methods have been developed for that purpose. Think-aloud and verbal protocol analysis (Ericson & Simon, 1993) and various advanced methods to apply them have been widely used. However, it is clear now that in the human case, this is far from a simple problem, and often it is difficult if not impossible to identify the critical mental processes in individual or collective decision making and the reasons for such behaviours in general. Verbalisation of mental processes is difficult for the test subjects and for those analysing that data. This becomes problematic when procedural or tacit knowledge is involved, when social interaction occurs, and when intuitive reasoning is used. It is likely that with increasing complexity of DL and other systems like GANs (Generatve Adversarial Nertworks) the problem explainability will be no less challenging than it is in the case of human decision making. We can expect the human and AI research fields to feed each other.
Solving the xAI problem
In the medical domain, for example, Holzinger at al (2019) look at the xAI problem from peer-to-peer perspective, as it occurs between medical professionals. The idea is simply that xAI should be able to conduct professional negotiations, man to man, woman to women. They present an excellent review of the black box problem as it concerns medical decision making in specific diagnostic situations (histopathology). Because of the hard diagnostic criteria and the risks involved in failures, the medical domain serves as an excellent case environment for building the theoretical and practical foundations for AI. Holzinger at al. emphasise the need for both explainability and causality of the AI systems. Large-scale industrial settings bear similarities with the medical domain – there the cost of failing AI can become high and complex process control and understanding is necessary. Both explainability and causality are needed.
The need for xAI is now widely shared where the complexity of AI environments increases. Costly false alarms, misses and unpredictable erroneous behaviours can be difficult to predict and trace in current systems. Extensive follow-up, time series analysis, off-line testing and continuous improvements are necessary to build successful AI based systems. On the other hand, AI implementation contexts very considerably and what can be tolerated in customer service or marketing can be a catastrophe in military, medical and industrial settings.
Gunning (2017) summarises the overall aims for xAI: a) to produce more explainable AI models and maintain a high level of learning performance (e.g., prediction accuracy and b) to make it possible for human users to understand, appropriately trust, and effectively manage the emerging generation of artificially intelligent partners. To accomplish this, Gunning emphasises that an explainable AI model needs an “explainable interface”, interaction and means of explaining the behaviour of the AI/ML system to a human operator.
The following are two example approaches in aiming at xAI:
The decision making situation can be replayed and discover which factors were used in each decision making situation and which were used when the situation changed (Johnson, 1994). Humans performing the same tasks can be used as a reference and try to find correspondence between the human and the artificial decision making processes.
It is possible to imitate human reasoning like in observing (visual) objects, e.g. cars or birds. To quote Chen at al (2019): “The model thus reasons in a way that is qualitatively similar to the way ornithologists, physicians, and others would explain to people on how to solve challenging image classification tasks.” Again the human specialists is taken as the reference.
At the moment there are several initiatives going on to solve the black box problem in AI and even an Explainable Machine Learning Challenge competition between Google, the Fair Isaac Corporation (FICO), and the academics at Berkeley, Oxford, Imperial, UC Irvine, and MIT, was arranged in 2018 (Rudin & Radin, 2019).
However, the field is still young, and the medical example case by Holzinger et al emphasises the need to make separation between explainability and causality. No doubt, this will be a crucial aspect in introducing understanding into future AI in large scale contexts. In real industrial settings this is a new challenge and to accomplish it without putting the AI-based processes at risk. Small steps can and must be taken now.
Human and collaboration approach to xAI
From the human side, the value of explanation depends on its clarity, usefulness, the way I facilitates the use of relevant mental models of the AI system and on the way its decision making performance and overall workings become understandable and even predictable within the working community. In general, an xAI system includes trust in how it behaves in variable situations and what it can, will and cannot do. The explanations offered to the users must carry information, knowledge and guidance of which the users learn something important or new, and which can be turned into relevant automatic, human-controlled or semi-automatic, well-defined actions.
These are important requirement of xAi, and the question arises how to accomplish this in real implementation of AI where a community and an AI implementation project work and interact? Is it only a matter of peer-to-peer professional negotiations?
Here we approach the black box problem from industrial-scale collaboration perspective, that is, we consider the interaction between the three parties – the human process operators and other site personnel, design specialists and the artificial AI/Ml system itself – interacting in the planning, implementation, using and tuning of the industrial scale AI/ML systems. This interactive “Triad” needs relevant information that is generated within this interacting entity: it does not originate from the AI/ML system or from the participating human resources alone but instead it emerges as an outcome of the triad interaction and collaboration. This is a new and acute knowledge creation paradigm in AI/ML settings. Because of the complex nature of this knowledge creation, it is good to consider what does “explainability” mean in such a collaborative contexts. We will not go into details here, but introduce some basic principles of collaborative xAI as we see them. We consider ‘explainability’ as collective knowledge acquired during planning, implementation and inter action with the system. From this perspective, it is not only a question of peer-to-peer negotiations but about collective knowledge building and sharing. Of course, peer-to-peer type of explainability can be an integral part of this.
Earlier we (Nyman and Kuittinen, 2019) described the “Triad of collaboration” where system designers, site engineers and operators and other personnell work together and interact with the AI/ML system to plan, implement, run and tune the system. Looking at xAI from this perspective reveals several “stakeholders” who need AI explanations from xAI in their work. We can discern the following instances where AI explanations are received, understood and used:
- Technological performance of the AI system is monitored by the design and engineering professionals who know the architecture and theory of the system in every possible detail. For them, there is the need for high-level of “peer-to-peer” negations (when one peer can be the AI/ML system), with some support from the specialists in the application field (operators).
- The domain of the specialist operators is the process (manufacturing, maintenance, service etc) and all of its parts where their task is to secure that the ML/AI system behaviour converges towards optimal performance in terms of product/service quality and quantity. Their knowledge domain is different from that of the design specialists. Hence, an “explanation” to them must carry relevant information about the domains of their responsibility and support them in initiating relevant actions and control.
- Management of AI/ML at large scale sites need information that includes elements different from typical automation and digitalisation programs. The black box problem, if not solved, introduces, for example, the following new management challenges: 1. How to gain efficient knowledge for deciding about significant investments into a full implementation of the AI/ML system and what kind of knowledge is critical in this? 2. How to follow the development of the system performance when it is being introduced and tested and how to extend it? Both of these needs include explanative knowledge that must be both technologically solid and understandable and as clearly as possible related to the operating domain of the site.
Explainability in different domains
We can return back to the explainability and how it depends on the domain of negations and knowledge needs.
Firstly, what is meant by “explainability” and “understanding” for the people working with and managing the AI/ML system? The main aim of xAI is to inform designers, operators, and the management to take action and to enjoy the positive feedback suggesting that the outcome of the AI/ML guided process is leading to the expected positive performance. Both situation occur, daily. However, many intervention actions that the operators must take require coordinated collaboration, often negotiations and reporting of what to do, how and when. Any explanatory information available for this, must be congruent with the ways of thinking, observing and negotiating the process that is being controlled. Hence, already form the start xAI must produce both AI-technical information and process related information.
The question then arises, how should the available xAI knowledge be formulated so that it can be used efficiently by the professional operators. Clearly, it necessary to facilitate collaborative knowledge creation which serves different knowledge needs. From the beginning the development of xAI must consider this. Simplified peer-to-peer consideration is not enough for supporting effective organisational decision making and actions.
We have earlier used the concept of ‘observational architecture’ to consider the critical information flows in AI guided environments. When building an xAI, it is necessary to consider this information or knowledge architecture and take the perspectives of different knowledge domains and needs. There is no unique solution to this since the information needs vary depending on the site and its overall environment. In essence this can be described as a demand to generate relevant actions from critical perceptions. Suitable models of collaboration, information flow, knowledge creation and action are needed.
The outcome of a mature, collaborative xAI is a knowledge community which shares relevant data obtained from xAI and represents it in different domains, in a way that serves the aims and purposes of the organisation or the site using AI/ML in its operations. In this sense, this new xAI knowledge is dynamic and it emerges and evolves with the improving performance of the AI/ML system. This does not happen without proper management of the collaboration within an xAI environment.
Chen, C., Li, O., Tao, C., Barnett, A.J., Su, J. & Rudin, C. (2019) This looks like that: Deep learning for interpretable image recognition. Advances in Neural Information Processing Systems 32 (NeurIPS)
Ericsson, K.A. , & Simon, H.A. ( 1993). Protocol analysis. Verbal reports as data (1st rev. ed.) . Cambridge, MA: MIT Press.
Gunning, D. (2017). Explainable Artificial Intelligence. DARPA/I20, Program update.
Holzinger, A:, Langs, G., Denk, H., Zatloukal, K. & Muller, H. (2019). Causability and explainability of artificial intelligence in medicine. WIREs, Vol 9, Issue 4.
Johnson, W. L. (1994) Agents that learn to explain themselves. AAAI-94 Proceedings, 1994.
Rudin, C. And Radin, J. (2019) Why are we using black box models in AI when we don’t need to? A lesson from and explainable competition. HDSR Nov 1/2019
February 4, 2019 § Leave a comment
I fed ”Life long learning” to our friend and enemy, Google and received about 3 100 000 hits. No doubt it’s a very popular concept and having such a massive popularity, one would assume it has a significant explanatory power in the study of human and especially adult learning and education. But then I tried ”Life long curiosity” and got only 26 000 hits. I got curious, why so few? You would think that curiosity is the golden force we envy from our children and which makes people move towards and beyond the frontiers of knowledge and human experiences.
When I first learned about the concept of Life Long Learning (LLL), sometime in 90’s it was a meeting at our Ministry of Education and Culture, I felt uneasy, but remained silent; the concept felt somehow artificial and fake, but I did not know why. Of course, we all learn from birth to death, some learn good things, and some learn bad stuff. Many seem to regress, undergoing a form of pathological learning, or why not call it negative learning. As sad as it sounds, most of us must learn to die.
There are numerous interesting-sounding concepts around LLL, like social software for LLL, learning any time and anywhere, learning infrastructure, empowerment, mobility, and many others, but when you take off the emperor’s clothes from “life long learning” you’ll be surprised to notice how, in real life, it simply means that even after school, university or any other educational institute, or when you become 30 or so, you must learn something new and useful. All the time.
I had an inspiring talk with a somewhat younger friend of mine about our ongoing, personal projects. We did not start talking about studying or learning, although formally and from the outset it was about different forms of learning we both had going on. It did not take long, before it was clear to us that it’s not about life long learning at all, it’s about Life Long Curiosity. We had curiosity projects going on.
Well, of course life long learning can be the outcome of a future educational system, and a valuable, if not a necessary one for us in the fast paced, future world, but something else must happen first, before we are ready and willing to dwell into learning, independent of the external pressures like the need for professional progress, new digital tools, the teachers, the lure of gamification or the pile of money promised.
But what is curiosity and how does it come about? Surprising enough, its’ not too well known. In his much-cited paper (1994) George Lowenstein defined curiosity:
”The new account interprets curiosity as a form of cognitively induced deprivation that arises from the perception of a gap in knowledge or understanding.”
According to Lowenstein it’s deprivation, like hunger or thirst, perhaps both! Typical to the spirit of the 90s the definition of curiosity, however, emphasizes its cognitive nature with a somewhat diluted sprit in it. We all know the tickling feeling when we are passionate to find out something compelling, good and why not even bad, so much that we forget everything else, even sleep; it’s much more than a cognitive gap in knowledge. Scientists who have studied such psychological phenomena have noted how curiosity drives, not only the work of the “doers” but also the reactions of people to arts, sports, sciences, education, advertisements, and to whatever.
When you follow the politicians – not only in Finland – they see gaps everywhere in our education system, its efficiency, usefulness, use of money, and strategic value. But the way they see the gap has nothing to do with curiosity – it’s more a question of immediate gain of power, fame and often it is simply greed. Curiosity and greed live in separate universes but sometimes, unfortunately, they overlap.
Modern educational scientists have been fascinated to adopt the concept of life long learning, and why not, it does offer a wonderful chance to extend the scientific and (sometimes) practical scope of their research and teaching to learning at any age. Politicians are happy to support that in the hope of extending the working years and to respond to what they call strategic needs of a nation. Researchers get more funding.
But of course, Lowenstein had something else on his mind when defining curiosity like that. He meant that an individual, a person, a human being perceives the gap and becomes curious about it and does something about it, like drinking from the well of knowledge. But a most fundamental question and challenge arises: what makes us, our grandparents, or our children perceive this gap of knowledge?
What I’d like to suggest here is, that curiosity as a psychological phenomenon is the most important source of mental, even spiritual energy in education and human growth. Curiosity makes us tick. Curiosity can take the adventurous mind to the limits of the unexplainable and to the most fascinating inventions of the mankind. Without curiosity, like without motivation, everything dies. It is now accident that “Curiosity” is the name of the famous Mars Rover (https://www.nasa.gov/mission_pages/msl/index.html) We should know everything about human curiosity.
Through some of the wonderful educational entrepreneurs in Finland and elsewhere we know that one of the underlying ideas in their ways of using games, methods, materials, and technological and educational tools to promote learning is to inspire and motivate students, to do it better than ever before. Often they are working with the power of curiosity although they do not call it that and instead use to the terms like inspire, fun, engage, innovate, and enthusiasm.
“Life long learning” is a benevolent descriptive concept to emphasize the need to learn from birth to death and to serve the society and ourselves, but it is not enough. Can you be curious if you don’t know anything? What kind of curiosity is based on a very thin knowledge background, or to express it more bluntly: Can an ignorant person be curious, too? What does excellent education level mean to the potential of curiosity? What is needed to make people motivated to learn?
In future we may well talk about curiosity technologies. We can already imagine and actually see it in the activities of the firms and initiatives like HundrEd (https://hundred.org/en) , Lightneer (http://www.lightneer.com), to mention only two promising Finnish initiatives.
One could imagine that Artificial Intelligence systems will learn to be curious about our world and us, but there is a mountain of problems before it can reach the level of human capability. One of the amazing aspect of human curiosity is the ability to change the domain of interest that can take place in seconds. In arts and increasingly more often in sciences it is a secret behind revolutionary ideas.
To finish with a personal experience on the power of curiosity, some 25 years ago I was observing how a researcher was using a text coding software (Atlas.ti) to analyze and code the texts from several interviews in her organizational research project. I knew nothing about the analysis method she was using, but was curious to know. I have a strong background in visual sciences and can even claim that at the time I knew every neural mechanism there was to know of the human visual system, especially from bottom up, from the retinal networks to thalamus to the visual and other cortices. I had nothing to do with that kind of analysis tools.
When she explained what she was doing – generating a kind of a pyramid of codes of the meanings in the texts – it struck me that why not study the visual system with a similar, rigorous, top-down approach, to study and model human visual experiences? I had spent all my research life studying mostly bottom-up visual mechanisms. We had just started a large project on visual quality and that curiosity turned my “pyramid of knowledge upside down”. The results during these decades have been just fascinating and something to be proud of. Without the misplaced curiosity – with respect to the ‘correct’ visual sciences at that time – I don’t think I had made the dramatic change.
There is a long way for the AI systems to achieve this kind of dynamic curiosity behavior and of course, it is not easy for us human either but when it happens, new doors can be opened by our curiosity, to new spaces for opportunity perception.
(Note: I published this post in Finnishnews/2017 but got again curious about the topic and republish it now here).
January 25, 2019 § Leave a comment
”Teatteridiplomi: Rakkauskirje Freudille
Superego, ego, id. Mitä kun alitajunta haluaa tehdä toisin, kun yliminä? Tunteet sanovat joo, mutta maalaisjärki kieltää. Päänsisäistä väittelyä sekä nuoren suhde-elämän luotsaamista kallonkutistajan tasolta.
24.01.2019 klo 18.00 Helsingin Suomalaisen Yhteiskoulun juhlasalissa, Isonnevantie 8
Kesto n. 30 min
Lavalla Alba Ala-Pietilä ja Julia von Lerber. VAPAA PÄÄSY.”
Yesterday 24th January, we attended “A love letter to Freud”, a theatre play written, produced and directed by Alba Ala-Pietilä and Julia von Lerber who both were on stage as well. The play was part of their studies at Suomalainen Yhteiskoulu lukio (high school) “Theatre diploma” where they both study.
It was such a wonderful experience that I wanted to share something of it here and perhaps, who knows, someone might invite them to perform. I loved the staging and the idea of the play so much that I don’t want to expose any of its details here.
Alba and Julia had come up with a fascinating way to represent and approach the internal conflicts and struggles we all experience between our drives, passions, motivations, emotions – and the rational self. In the invitation the young artists refer to Superego, Ego, Id and Freud, of course. But it’s about today and what a fresh and liberating way they had to make all this visible and tangible. There were no signs of destructive anxiety, no mania for positive psychological solutions, it was about enlightening conflicts of everyday mental life and ways to live and love in the middle of it all – as we all do, or could do. The play was even fun and educating to follow, and to feel the internal conflicts of the young mind on the stage – and for sure, of the older ones in the audience as I could verify myself. Note that I use the term “mind” here although there were two actors on the stage.
The play is probably aimed at a young audience but it is much more as I could experience when a cavalacade of images from my own teen age, even middle age and current life started entering my own stage of mind. I don’t remember many times that it has been so liberating and delightful to feel and re-experience them, the internal conflicts that have the label ‘problems’. Without spoiling the potential joy of seeing the play (I hope it will be on Youtube or elsewhere), here are some reasons for my enthusiasm about what I saw.
From the start the stage span a space which could be from any famous, intimate theatre house In Europe, it was like a beginning for a classic Italian or French movie. The actors acted as a puzzle to the audience who slowly (at least I was slow to realize it) sensed the core secret of the play: the way to represent the internal, mental struggles and show its consequences and dynamics during the play. What a charming and kind way these young women had in their way to lead the audience to the gates of consciousness. The actors were simply wonderful in creating this scene and its inviting atmosphere. I could feel the defenses shatter in the audience.
Immersing into the play I started to realize there is no psychological limit to how far into a troubled consciousness their play could take us and I even felt the first signs of worry or anxiety, expecting – perhaps – painful conflicts and disasters of life to crowd the stage. But even that was accomplished in a kind and caring way and the audience was free to enjoy the personal developments. This reminded me from some of the plays directed by professor Pauliina Hulkko, a writer and dramaturgist, who is a master in inviting her audiences to the edge of serious questions of life while at the same time caring for their well-being and motivation, making sure they want to come closer to even extremely difficult themes.
The final scene in the play was full of internal light, joy and promise, a future for the young souls and reminder of the wonderful history of the futures of us all, the older audience. One thing was certain: the mental struggles will continue, forever, but it need not be suffering only. Never have I heard the sound of “You have a message” been so well placed and liberating. It made me smile and laugh from the heart.
If they are now observant at the school they will invite these skilled, young artists to do the play again after ten years or so, when they have their own, new experiences of life and living behind, to adopt the text to their new situation. The stage is there already; it is ready for anything and needs no transformation. What a wonderful opportunity!
October 1, 2018 § 2 Comments
Göte Nyman, University of Helsinki,
Ossi Kuittinen, SimAnalytics Ltd
In traditional industrial and service environments, an ideal AI system is meant to reach the competence level of human operators and their teams, and then outperform them as soon as it is technically, economically and resource-wise possible and can be managed. Inevitably, any organization adopting AI will meet – it is occurring all the time – its first critical moment when this happens, for the first time. For firms, services and industrial settings this significant event is a wake-up call, both to the whole organization but also to its AI developers.
Photo by ICAPlants – Own work, CC BY-SA 3.0, Wikimedia commons.
The learning AI system becomes accurate, fast and excellent in securing production quality, offering failure diagnosis and predicting service outcomes. It is perhaps taking its first steps in predicting, managing and solving more complex problems than has been possible before by human operators and other professionals. The criteria of these significant events involve several (socio-technological) variables and it is not self-evident at all how to recognize them and how an organization should react to them. We should understand well these first moments where AI it appears to outperform the operators. Here we explain why.
What is happens when the AI system outperforms its teachers?
This special, but very practical step has not been a popular discussion topic while the focus has been on potential future risks, job transformations and even disasters when AI becomes strong enough. However, it is a new window of opportunity for the organization and its people and should be recognized as such: if a firm or a production unit makes mistakes in interpreting what it means and how to respond to the new situation it runs at risk of missing the further development of the AI based work and processes. It is not only an alerting signal to the technology use but to human resource management as well. AI does not know where it leads the working community, it does not ask and it does not care.
The personnel responsible for the ‘old’ system and implementing the AI can face uncertain times. This is not a new phenomenon in businesses and industries experiencing digital transformations; similar situations occur in banks and insurance companies, for example, when they change to more modern technological solutions while it is necessary to run the old systems. The same has occurred in major firms in Silicon Valley as well, where the personnel knows their value on the job market depends on their ability to work with the latest technology. Investing time and energy in old systems is risky.
Division of labor, team and management functions, will all become under reconsideration. Soon, also the pay and reward systems must be aligned so that they support the new AI-based situation at the site: traditional pay models may not work any longer and can be even harmful if they reward for wrong activities and miss the critical ones. Before this renovation can be accomplished it is necessary to identify the major factors contributing to the system performance.
One could call this new phase a strange form of ‘organizational interregnum’, when the organization is aware that its process control must be changed but it is not clear yet, what should be done and how to run the two systems in parallel; the change cannot be accomplished overnight and its necessary to keenly follow the performance of the AI system. It may well take more than a year depending on the scale and nature of the AI implementation. During this time, the organization, its management and the personnel in general must build trust on two frontiers: trust in the implemented AI technology and its use and trust in the collaboration among the personnel, teams and management. Failure in either of these leads to uncertainty, loss of motivation, conflicts and hinders in decision making.
Paradoxical as it may seem, when AI reaches its first ambitious goal it will need the best human knowledge and competences and new ways of working together, new teams, for example, just as it had needed them during the implementation, when starting to learn its new tasks. If handled well this can become a moment of growth and inspiration to the personnel, a chance to build an effective work environment. Adoption of AI is a challenge to people committed to creating a healthy and fair workplace, to learn new required skills and to secure a good performance. It is the responsibility of the management to provide conditions for this.
Management’s task is to secure both social and situational awareness in the company, to facilitate seamless collaboration and mutual understanding. Facing the unpredictable future of the fresh AI, the best strategy with personnel is to be proactive, especially in education, and to build a realistic image of the expected development.
For a factory, firm or industrial plant it is a matter of tuning anew what we call “observational architecture”: finding out where are the most important information sources of the site and its AI system, what phenomena to follow, and how to report and act on them. The architecture must include the personnel, too. This is nothing new for the management: healthy team work, collaboration, learning and communication are needed. However, their content and form are changing. Furthermore, there is an acute risk: if misused or having failed to see design faults in the AI system, serious problems, of a magnitude worse than ever before, can follow. The risks must be identified during the first implementation phase. This requires efficient participation of the personnel so that early warning signs are taken seriously.
How to communicate with an AI companion?
When AI performs better than the operators used to do, what knowledge has it acquired and how could we find out what is the essence of this new machine competence? It has learned “intelligent” input-output relationships that humans have traditionally mastered but it has learned something else, too. “What might that be?” is a question that will be repeatedly asked when AI is adopted. Interestingly, the above question is not far from the classical problem of behaviorism in psychology: should we focus on the observable behavior only or should we try to see what happens inside the “box”, what underlies behind its behavior and decision making? AI and especially the deep learning systems have reached a satisfactory black box performance and now its designers try to progress from that, to understand what happens inside these systems. However, the fellow in the AI box remains known.
In his MIT Technology Review (April/2017) column, Will Knigth starts the story “The Dark Secret at the Heart of AI”, with an ingress. “No one really knows how the most advanced algorithms do what they do. That could be a problem.” As an example, he uses the self-driving car designed by NVIDIA, which has used deep learning to observe how humans do it and learn by observing. Knight then asks: “But what if one day it did something unexpected—crashed into a tree, or sat at a green light? As things stand now, it might be difficult to find out why.”
With self-driving cars, this “dark learning” can be a multi-layered problem extending from simple pattern recognition aspects to traffic safety and to ethics of decision making. However, less often we hear about similar problems from industrial plants, service providers or other organizations launching the use of strong AI to run their processes.
We don’t know how to best talk to AI systems and how they should talk to us. We don’t know what new and valuable knowledge the best AI systems would have to tell us – if they had a means to do it and we could understand it. It is no longer a distant philosophical question but a very practical one as we have seen in complex industrial settings. AI can be taught to follow instructions and complex rule sets and even have a reasonable sounding conversation with us, but it becomes more difficult to communicate with it, especially in case of deep learning, running large-scale processes, The MIT story reminds us that it’s is about building trust in the AI system – but it is essentially human trust, something similar we experience towards our cars – and tools are needed for it.
It is easy to see when an AI system outperforms human operators in speed, accuracy and the quality of output in a specific task. Standard methods and metrics cover these basic measures; interpretation of the performance data may not be that simple: why was the AI system able to perform so well? Did it use the same information as the human operators have always used, but more efficiently or did it find its own ways of deriving new knowledge from what it had learned during its teaching/learning process? We should know if it has learned potential, dormant behaviors – which have not yet occurred – that are harmful for the system. Curious enough, many technologies show implicit trust in its users: their design assumes that the users do not act in certain (dangerous or stupid) ways even though they could easily do that.
First, second and third order learning phases of AI in industrial practice
When an AI system is taught by feeding it with offline data and then later real-time data from a real process, it is learning to behave as expected and to produce the hoped for beneficial input-output reactions. We call this first order learning of AI, when it still does not outperform the ‘old’ practices. The operators can compare the system performance data – as it has occurred under manual or semi-manual control – against the data obtained with the AI system running. Based on this comparison the AI system can be tuned, given additional teaching material, improve the quality of input data, and identify any needs for additional data sources.
This is a relatively complex, socio-technical development phase where the immediate aim is to guide the AI system so that it could match and eventually outperforms humans. When it is implemented for the first time at the site, it is natural to follow and rely on the same performance metrics and other work practices that have been used when the process was controlled by human operators.
Because of the huge data output capacity of the AI system, specific user interfaces (UI) are needed, with the capacity to show and allow safe control of its functions and performance. New data representations and system controls must be used and tested. However, there is no unique standard to define what the output of an AI controlled system should now look like and what would be its core purpose. From a running AI system, it is possible to have real-time and computed data from thousands of measurement points which it must show to the operators whose task is to interpret the flow of information and evaluate the system performance. But of course, humans cannot follow so much data. Effective representations, condensed, packed and informative, must be invented, just like in nuclear plants, for example, but what should such an UI be like and how should it be used?
It is our belief that the UI:s of AI will evolve fast and new concepts and forms will be continuously introduced to serve the specific AI contexts. They will have a crucial, even competitive role in supporting human work and collaboration and in helping the personnel to understand the new control environment. Investing time and resources in wise UI can have a major impact on the system performance. There are good grounds to take this very seriously, especially at large scale industrial and production sites. Often when AI is introduced it touches a major part of the organization. During the first order learning phase the initial UIs are designed on site, but the need to improve them becomes quickly apparent.
The second order learning phase has a special nature since it is the moment when the AI system, for the first time, outperforms human operators. This becomes a new high-level challenge: how and why did the AI system reach such a good performance? What data and controls were most informative for it and how should its performance now be described and represented? How should this new AI knowledge be presented to the operators? What kind of representations are best for describing the important states and performance of AI and the processes it controls? Can the system offer guidance to the site personnel about how to improve the process and the infrastructure? How should the operators be offered ways of controlling the AI-based processes? Is it at all reasonable to consider the interaction with the operators and AI as a UI problem or is some other approach needed?
This is only the beginning of a major transformation in the AI-Humans interaction and touches the participating designers, engineers, operators and AI specialists. We have earlier described a triad model of learning (Nyman & Kuittinen, 2018) where the operators, designers and those responsible for the engineering and implementation of the AI system, work together, learning from each other and from the AI system. Now this situation is at hand again, but the context has profoundly changed: the AI system does what it was intended to do, outperforming the human operators. What does this mean to the triad community, what should be done next and how to prepare for its future?
This is the start of the third order learning phase where the learning extends to the organization at large. It becomes a major organizational and management task – to reorganize under the pressure of the emerging, new needs and possibilities. The reason to this is simple: an effective AI system introduces new constraints on the industrial plant or a service provider, for example, which must now re-organize its work, modify its organizational structure, and secure the new competences for building the processes around the running AI system. It can be a time to rethink the business model as well.
During this race for performance and quality it is often forgotten that the organization must be prepared to change for a better AI technology when it is useful and possible. Finally, with the fast advancement of AI we must make sure that we don’t end up becoming the victims of it but instead can co-work with AI to secure our good life and work with human purpose.
April 23, 2018 § Leave a comment
Göte Nyman, University of Helsinki, Finland
Ossi Kuittinen, SimAnalytics Ltd Helsinki, Finland
Machine Learning (ML) as part of AI is now familiar to us all, from managers and workers to engineers and designers. We know the secret of ML: it’s amazing potential to learn from data, huge amounts of it, which sometimes is a guided process but can also proceed without significant human intervention during the learning process. However, human contribution and collaboration remain necessary for any ML system design and implementation and especially in securing optimal ML system learning and its continuous improvement.
(Creative Commons licence image)
What is the process where skilled human professionals and a ML system join their competences for a learning journey? How should we characterize the teaching, learning and interaction, where the ML system, human designers, engineers and workers act and learn together to come up with the best possible ML solution to the target environment?
With the fast improving ML we are suddenly confronted with a new kind of a teaching-learning-collaboration challenge or paradigm, where the competences, roles, learning processes, and responsibilities of the participating people must find their effective form and content, together with the ML system.
One could think that human-technology problems are nothing new and that the introduction of ML is just one phase like any other major leap in technology. But it may come as a surprise that we face a new kind of an interactive learning paradigm which changes fast. We call the work with ML as ‘triad of learning and collaboration’, consisting of at least the following components/actors (We use the concept of ‘triad’ in a slightly transformed form as borrowed from the learning approach where ‘the individual’, ‘the learning community’, and ‘the authentic use of the objects’ are separated entities of the learning situation; Paavola & Hakkarainen, 2005). A ML system that learns, interacts and teaches is a complex creature, indeed and this makes all the difference:
- The ML system as a whole – its sw, hw, system integration, computational models used, interfacing and its UI
- Operator-professionals specialized in the target domain and facing the planning, design and implementation of the ML system
- Designers and engineers responsible for the ML system design, implementation, process quality, and its continuous tuning.
What is special in this triad, is that all three ‘partners’ interact, teach and learn from each other, from the beginning of the ML project, often simultaneously. Furthermore, while there are a plethora of theories and models of learning, including computer-assisted learning, for example, the theory of very intelligent human-ML-system, interactive learning – where a human teaches a machine and the machine teaches humans – does not exists. It is time to launch its development so that it can keep up with the pace of fast advancing machine learning. There is already an intense discussion on the possibility to look inside a working deep neural network, one of the ML application, to improve its interaction with its operators but this only starting. Furthermore, it is highly likely that due to theoretical and technological advances, the current AI models will undergo major changes within the next decade or so.
We have summarized the triad learning relationships in a simplified form in the figure, where the instances of ML-human learning occur: 1. between the User/Operator and the ML system, and 2. between the Designer/Engineer and the ML system. The two instances are very different in nature and require relevant organizing and management, because of the differences in the human competences involved. Both learning processes can occur over the whole development and maintenance cycle of the ML system, with different weights over time and they are of utmost value to the success of the ML implementation.
Figure. A simplified scheme of the learning/teaching relationships and some of the learning contents in the ML implementation.
Design and introduction of the ML system
The design and project plan for the implementation of the ML can best be described as co-design (some form of it, from weak to strong). The designers/engineers introduce the candidate ML model for the management of the target system – an organization, industrial plant or other relatively large-scale enterprise where the benefits of ML are expected to cover the costs. This is a complex learning challenge requiring the participation of the designers and relevant specialists who know and work with the target system and know the processes and who will then work with the ML-system. It is dangerous to apply a linear and fast-to-the-implementation approach in such a complex endeavor.
While the mathematical and technological features can be too complex for all users/operators to follow, it is necessary to introduce the ML in a way that its basic learning and its teaching characteristics can be understood and that a fruitful interaction in co-design becomes possible. In other words, a valid, ML model-in-practice must be introduced to the participant operators and other personnel. Usually there is no unique or single ML model available for this purpose; hence the introduced model must be formulated and tailored accordingly, to match the competences of the participating personnel and the specific context of ML implementation. At the beginning, it is important to relate the ML concept, its functions and measures as closely to the target system model ‘as-is-model’ as possible so that the participating people can relate the ML system functions and its measures, the metrics used and the UI to the target system with which they are high-class professionals.
During piloting, testing and implementing these new concepts become critical since they make up the language with which the operators and users ‘negotiate the system’ and try to understand how it works. The operators learn a new ‘mental model’ of it by which they know what the ML system can and cannot do, what risks are included, how it can best be taught and how to communicate about it with the designers. These concepts guide the operators in their observations, too: what to follow, what to search for and how to react to certain observations of the ML system state and behavior. This is a most demanding challenge and mistakes made and vague concepts introduced can lead to blind spots in the development project, valuable knowledge of the ML system performance lost, underperformance and the costs becoming high. It is a very serious learning task for both the ML system providers and the personnel acquiring it.
New paradigm of interaction:people teaching ML when ML teaches people
With its computational power the ML system can shoot out practically unlimited amount of data: new measures if its status and dynamics, holistic and specific component state information, and any aspect of its measured performance, fast and slow in nature, depending on the environment. Overflow of information is a real practical risk and possibility and it must be dealt with by means design, and human learning and collaboration. Some of the ML data is totally new to the people specialized in the ‘old’ system having its own, perhaps historical control systems and performance metrics.
Some of the ML data can be significant and even crucial for the system performance, some not, and during the piloting, testing and implementation process the personnel learns to recognize what is relevant information and why, and then guide the designers in representing it in the UI. This data environment then becomes the de facto world where the personnel learns to work – with the (partly or totally) new information.
The ML data must be made accessible for easy observations and its background processes amenable for easy, immediate and relevant control whenever necessary. An explicit interaction loop (measure-control-measure) must be provided which is either automatically controlled by the ML system, and only made visible to the operators or which is amenable for human control. The observation of these (numerous) control loops is an essential part of human-ML learning and it will introduce new tasks, practices and responsibilities to the personnel.
What is new in this triad-entity is that the ML system teaches both the Operators/Users and the Designers/Engineers in various ways, some of which are new. How does this interaction take place and how should this learning be considered and conceptualized as it concerns both the personnel and the ML system? This is a fascinating learning theoretical challenge which cannot be dealt here with now, but we will return to it later, especially after having collected practical experiences again from the work with SimAnalytics Finland.
February 5, 2018 § 1 Comment
This is may be a speculative blog from the technological perspective but its behavioral background is solid; at least I believe so. The idea presented here is somehow funny and very serious at the same time. The simple question I have on my mind is how to teach manners to AI? It is not about a polite AI only, the problem scales up to as high spheres of human behavior and culture as far as we can see.
Seeing only a gloomy AI future?
We have read and heard Elon Musk and Stephen Hawking painting a scary future for the potentially destructive AI if it manages to escape our control and starts running wild. Some may think that ‘we can always take the plug off’ or that ‘AI has no will’. However, observing the recent false missile attack alarm in Hawaii made it clear to everyone, how simple human errors in using technology can cause devastating effects. It was a wake-up call to me too, especially noticing the official reaction to the error. The person who made the mistake was fired – a strange solution to the catastrophe – but the designers of the UI run free, I assume.
In AI the risks can become much worse than in Hawaii, especially when human errors can trigger complex, and difficult to follow chains of AI-based actions and the design for human-technology relationships is unfit to prevent this. The Hawaii example was extremely simple: the operator chose a wrong function from a simple, easily understandable set of alternatives. The mistake was taken seriously in the relevant organizations but I have not noticed this incident to launch much discussion on how we should prepare for the coming of AI where similar ‘human errors’ (actually they are design errors) will become possible.
In a recent panel discussion at Fire 2017 (Future of Work for Humans and Machines) the participants Joseph Smarr from Google and David Brin from Future Unlimited seemed to agree that it takes some time, some years perhaps, before the risk of a AI running dangerously wild becomes real. However, they did not discuss the ways in which AI could start living its own dangerous life in the net already today. Brin did imagine an AI, for example, which would be able and have a chance to scan and look at all the movies there are in the net and to learn whatever human behaviors we know is available there. What it would learn form the movies would not always be the best of humanity, so we need to find out human-controlled ways to teach the future AI manners and good behaviors. We should start it today.
Making AI a better person
We must teach AI manners. It is not different from educating our children and showing them how to behave in different life situations. At the moment there is no unique and scalable way to achieve this for AI.
Following the FiRe 2017 panel discussion and some of the comments from the audience made me think about the following: how to teach AI such manners and to do good or as someone from the audience suggested, to even nudge us to be better humans? AI cannot do anything like this unless it has a chance to learn behaviors that are good in nature, in some agreed-upon, human sense.
I’ve earlier introduced the concept of Internet of Behaviors where the idea is to introduce individual behavior data into the net and to make it (globally) available for a number of purposes, from health care, entertainment and education to marketing. The psychological thinking behind IoB is described here. It is like IoT but the idea is to assign addresses to an ongoing (it can be a historical or fictional, as well) behavior, which makes it possible to address and follow such a behavior and everything physical, digital and virtual related to it. This would also make it possible to build a contact with the person X showing that specific behavior (in case he/she is willing to allow it; I will not deal with the privacy issue here).
What if there was a systematic way to offer models of good behavior for AI to follow, to teach it behaviors we know and define as good behavior? In many cases it would be easy to define the criteria and to use such behaviors as models for AI to follow and learn. With the Internet of good Behaviors (IogB) approach we could offer AI access to behaviors (and companionships J ) we think are good for its development just like we do to our children. By allowing this we would let it use all the relevant data related to that behavior and to learn from it. It is quite possible we could learn from that too, but that’s another matter.
Of course there is no IogB system as of yet, but the potential exists already and is in use where personal data is collected by various devices. We do know how the deep neural nets already learn from examples but it’s some way to teaching them manners. IogB would take their learning to a higher level.
My idea for the AI community is to start a trial within a well-defined AI context where we know the criteria for good behavior and where good manners are relevant. It can be as simple as being polite in certain cultural situations, ways of speaking and interacting or getting food and support for the poor and looking at various ways people and citizens are now doing it globally, helping those in trouble, all over the world. Only imagination sets the limits here.
To run such a trial, we should arrange for people to adopt a coding (addressing) system for their detailed behaviors by using a simple app and monitoring system. Of course they should be willing to reveal (without revealing their identities) their behavior for the specific AI we want to teach manners. A feasible coding system for such behaviors is needed; you can consider this as a process of addressing specific behaviors in the same way as objects are addressed in IoT, which can be, for example, verbal or bodily expressions, emotional states, but they can also be physical or virtual transactions relevant to the specific entity of good behavior, practically anything related to human internal or external behavior. My main point here is that the occurrence of the addressed behavior makes it available in the net and the AI can then use it as learning data. There is much to do in this and to build a Teach manners to AI framework.
We can imagine the huge scope and scale of the approach by considering all possible contexts of good human behavior, from documents, and movies to real, human ongoing behaviors. Then there is the scary thing: Internet of bad Behaviors. It’s possible that we cannot stop it unless we can teach good behaviors first and even that may not be enough. Without going deeper into this I see a real, even important possibility for building and educating a human AI. We have time to do this. In the IoB blogs I have explained the background of this behavioral approach in more detail.