MIT professor Eric So makes a point about AI that I think is worth sitting with. In his interview with MIT News, he describes how easy it is to get a finished answer while missing the practice that helps us understand it.
I’m excited about what AI lets us do. I use it early when I’m working through a question, and I want more people to have that opportunity. But I also want to come away knowing more than I did when I started.
There’s work I’d gladly hand over. Moving the same information between systems or fixing the formatting on a document doesn’t need to take half the afternoon. Understanding why a decision makes sense is different. We need time to ask questions, try something, get it wrong and figure out what we missed.
So also describes using AI as a tutor. That’s the part I’d like us to put to work. Ask it to explain something another way. Have it question your reasoning. Work through an example together, then try the next one yourself. Getting help can make learning possible when you would’ve been stuck otherwise.
For me, asking AI first still means bringing my own judgment and worldview to what comes back. I need to be able to explain why I agree, where I disagree and what I’m going to do with it.
If we’re encouraging people to use AI at work, let’s give them room to learn with it. A useful question after someone finishes an assignment is, “Can you walk me through how you got there?” I’d want us to make that conversation part of the work, especially for someone who’s still learning it.