6 SCHWERPUNKT_INTERVIEW PERSONALquarterly 02 / 23 Teams are formed in the working world in a wide variety of areas. Working together with others is a natural way for most people to achieve common goals. For example, synergies can be exploited from the different competencies of the members. More and more teams are using artificial intelligence (AI) for decision making and automation. This raises questions about collaboration between humans and artificial intelligence. PERSONALquarterly: How do you collaborate with AI in your day-today work? Leslie DeChurch: At the moment I use it in pretty silly ways actually. I have a Google Home in my office and I often ask it about my next meeting, or to schedule reminders, do quick and easy math, or look things up. Mostly I love having the ability to talk to it, that really changes the feeling that you are collaborating. PERSONALquarterly: How do teams in different industries collaborate with AI? Do you have examples in which AI is more than just an assistance system? Leslie DeChurch: There are so many different industries that are starting to use AI in different ways. I think some of the early examples of AI were helping people make complex decisions. We all know the famous example of IBM‘s Watson beating jeopardy super champions on live TV. That technology has since been deployed in settings like healthcare, helping physicians diagnose complex cases. AI has been extremely useful in many data-intensive fields. Some of the most interesting to me personally have been the ones in the creative industries. New software applications help designers to create things. I was at a recent academic talk where all the illustrations and artwork in the slides were generated by AI. Your question about more than an assistant system is really an important one. I think that up until now AI really has been in an ancillary role assisting human beings. I‘ve gotten very interested in AI from a research perspective because increasingly there are use cases where AI joins as a full-fledged teammate. You can imagine military applications where AI is embedded in sensors and vehicles embedded with human soldiers. We can also imagine AI making recommendations to physicians in mediAdding AI to human teams can really change the social dynamics between human beings Das Interview mit Leslie DeChurch führte Simone Kauffeld cal and healthcare settings. We can imagine AI playing a quasi leadership role in important discussions detecting patterns in information sharing and then making recommendations about critical questions to consider. In all of these applications the use of AI is more than an assistant and rather it‘s something that is fundamentally changing the dynamics of people interacting with each other, assessing the value of one another‘s information and decisions. There are interesting social dynamics that could come about. I run some simple experiments in my classes where I have teams doing basic problem solving and creative thinking tasks and I have them do it over zoom with the technology that they think is an AI, but it‘s actually a person behind an avatar. During the debrief I am struck by some of the students comments where they will reflect on feeling hurt or excluded from the conversation because one of their human teammates preferred the ideas of the AI. Or at least what they thought at the time was an AI. There will also be cases where one member of the group gained status quickly because of their ability to interact proactively and to get more detailed information from the AI. These are some simple illustrations of how adding AI to human teams can really change the social dynamics between human beings and how they intuit who has valuable information or who they can trust or who‘s motivated. AI as a teammate can reshape a team’s status hierarchy. PERSONALquarterly: What are possible areas of application for hybrid teams? What is the difference between purely human teams, teams working with AI as an assistance system, and hybrid teams where AI becomes the teammate? Leslie DeChurch: To me the real critical difference between AI as an assistant versus a teammate comes into play when the AI starts to have agency and motivation. When we think about the example of an intelligent assistant, it waits to listen and respond until you specifically address it and ask for a particular thing. The AI doesn‘t have any kind of programming to be acting on its own beyond the bounds of what the person is telling it to do. But when we think toward future applications, we can easily imagine cases where certain motivations are embedded in the AI, and as it learns it exhibits its own intentions without merely responding to assist a person.
RkJQdWJsaXNoZXIy Mjc4MQ==