This article modified and developed an article published on “ER” No. 4 (pp. 39-40) in Japanese on 10 January 2017.
The relationship between humans and artificial intelligence (AI) raises the question of how we want to engage with others and work in society. There is no single answer to this question. Therefore, it is necessary to deconstruct the question and use trial and error to answer it. In this article, I will introduce a method to generate questions to visualize a society that coexists with AI, and create new values through dialogue. Instead of a passive view wherein work and life are taken over by AI, an active view of how you want to co-create with AI is important.
The workshop should have diverse participants.
It should be designed as a dialogue to listen to each other, rather than a discussion to emphasize opinions. Therefore, rules such as “do not criticize,” “do not be afraid to change the subject” and “call each other by name and not by job title” are necessary.
Each person chooses themes such as “life” or “work” and writes the tasks one does from waking up until going to bed on sticky notes.
The list will have various categories and content.
E.g. Live: Brushing teeth, taking showers, commuting
E.g. Work: Compose chats, experiments, discussion with your boss
It is useful to adapt the rule of “building on other people’s ideas.” This brainstorming session can take about five minutes.
Divide people into groups of five to six.
Draw a cross on the paper.
On the vertical axis, write “Tasks you want to leave to the machine” and “Tasks you don’t want to leave to the machine.”
On the horizontal axis, write “Technically possible, probably within 10 years” and “Technically impossible/unknown.”
Draw a circle in the center of the cross and writes “Tasks not reached a consensus” in it.
Each people chooses the tasks they want to allocate to the AI. However, if the group fails to reach an agreement about a task, then the task might be too big or ambiguous in interpretation. If so, the group will discuss whether that task can be further divided.
For example, “shopping” can be divided into “shopping for hobbies and stress relief” and “supplement of daily necessities.” In this case, the former can be divided into “Tasks you don’t want to leave to the machine” and the latter into “Tasks you want to leave to the machine.”
However, the goal of the workshop is not to reach consensus, rather, to talk about why consensus could not be reached. If an agreement cannot be reached, then pin the tasks to the middle circle and move on to the next tasks. Take around 20 minutes to sort out the tasks that each person writes on the paper.
Tasks you want to leave to machine
I noticed an interesting phenomenon in the workshops I conducted: many of the tasks were unexpectedly allocated to “Tasks you want to leave to the machine.”
Before the workshop, many people were worried about allocating the tasks to machines, however, during the workshop, they realized that there were many tasks that they wanted to leave to the machines.
Hence, the next question raised is: why were those tasks not done by by machines yet? This may be because of institutional and habitual barriers, or the complexity of the tasks. The kind of tasks that humans do make us wonder whether complex, exceptional, and troublesome tasks will be left to us, while machines take over the tasks that are easy and comfortable.
Tasks you do not want to leave to machine
If we look at the tasks that we do not want to leave to the machine, we can see the diverse values of people.
When I conducted a workshop with people from companies, many thought that humans should attend to their clients’ requests, even if that task could be done by machines. Although it might be technically possible, building trust with people is another issue. Reliance on machines might depend on how one wants to build relationships with people.
When I conducted a workshop with university students, some said that they wanted to do tasks that are related to their body, such as “taking a bath” “brushing teeth” and “muscle-training.” People who did not want to rely on machines said that the process of muscle training is important.
When I conducted a workshop with the elderly, some people allocated “bed making” and “cleaning rooms” as tasks that they wanted to do by themselves because it helped them feel confident about their ability to move their bodies. In general, we tend to think that we should leave these kind of tasks to machines, but to some elderly people, these are important part of their daily routines.
The workshop will help understand each person’s preferences and values, and provide not only superficial support but the kind that goes back to the basics.
Reasons for allocating the tasks
It is important to differentiate between the tasks that “one really want to do it by themselves, but leaves it to the machine because it reduces time and it is efficient” and “one does not want to do even if there is time, and hence allocates it to the machine.” These two tasks are qualitatively different. It is mistaking the cause for the end result by allocating the first task to the machine.
For example, in a workshop at a company, there was a comment that said: “I would like to take care of children by myself in the morning, but I do not have time. Therefore, it would be helpful if the machine could do that for me.” In this case, a system could be put in place wherein the other tasks that were taking up her time could be done by machines and other people. However, if people consider only the last sentence of the comment, then they might think there is a need for a machine that takes care of children. In fact, such a system is starting to emerge, but it is necessary to go back to the original premise and reflect on whether we really want to build such a society.
Task dividing abilities
After the workshop, people might realize that the ability to divide tasks imaginatively is the skill needed in the future. If one could divide a task into appropriate tasks, people could reach a consensus. For researchers, this could mean a new approach to study different topics. For making policies, this could provide a starting point for coordination toward consensus building. For businesses, this could be an opportunity to brainstorm ideas for a new business.
Limitations of dividing the tasks
Finally, we should be careful about splitting the tasks. There might be an important implication that you might miss. For example, can we break things like “apology” and “marriage” into tasks, or does that change the meaning of the original task? If we can, what values will they be rooted in?
Dividing and classifying tasks in this workshop will create new questions. Why not create questions and talk about co-creating work and life with AI?
Written by Arisa Ema