Do you often use AI chatbots such as ChatGPT, Claude, CoPilot, or Gemini to brainstorm for you? How effective do you find this process? Kellogg Professor Brian Uzzi ran an experiment with his students to examine the efficacy of these models and student attitudes about them. Uzzi administered the Divergent Aptitude Test (DAT) to his students. In four minutes, they had to generate a list of ten words that were as different as possible from one another. He compared the students' creativity to the AI models.
Uzzi found that the humans produced more unexpected responses than the AI models. In other words, creativity flourished among humans, while the AI models produced average answers. Unfortunately, Uzzi discovered that the students tended to prefer the responses generated by the AI chatbots. Why? Simply put, the students are impressed by the efficiency of these models. Uzzi says, "They get sucked in by the efficiency. Someone in class will say, ‘The bot’s score is no better than mine, but I get it in 10 seconds instead of several minutes.’ To them, that feels like a good trade-off.”
Uzzi doesn't believe that this simple experiment argues against all use of AI models. Instead, he argues that we have to think carefully about HOW we use the models. He explains, "To get the most out of a bot, don’t ask it for answers. Ask how to approach a problem. You want advice on how to think, not what to think.” In short, don't ask the AI model to replace your original thinking. Ask it to complement or supplement your way of working.
Let me offer an example from my use of AI to develop teaching plans and materials. I don't simply ask one of the chatbots how to teach a certain topic. Instead, I tend to think creatively about how to teach a subject, drawing on my years of experience as a faculty member. Then, I will ask AI to help me develop certain teaching materials that will achieve my goals. Thus, I'm engaging in the creative act, while asking AI to help with the implementation of my ideas. In this way, I'm combining human creativity and AI efficiency.

No comments:
Post a Comment