AI has made it deceptively easy to turn out polished work without engaging in deep thinking, and MIT Sloan professor Eric So looks at the cost from the employer’s side. He often sees people generate something impressive with AI and then fail to explain it at all. When there is a problem with it, they cannot fix it, and that, he says, is not really useful to him as an employer.
So made the case in a talk for “AI + X: How AI is Changing the Management Practice,” an MIT Sloan speaker series on how AI is shaping real management decisions. The series, co-sponsored by the school’s Office of Teaching and Learning and MIT Open Learning, takes a practical, cross-industry view. His point is about what sits behind the output. Employees need to understand the process, the reasoning and the ideas that produced a piece of work. Those are the skills employers value, and they are the skills that can weaken when people outsource too much of their thinking to AI.
The weakening is hard to see while it happens, because the output still looks polished. It shows when someone asks the author to walk through the work, or when the work breaks and needs fixing. A person who reached the result through their own reasoning can do both. A person who only reached the output cannot, however good the output looked.
So puts weight on valuing who you are without AI, and he sees that mentality as the reason people need to be taught differently. What counts is the process, and what a person develops along the way, because that is what lets them explain the work and repair it.
His new book, “The Collision: What AI Does to Us,” takes the argument further, into how AI is reshaping human cognition. It looks at how people can use AI to strengthen their critical thinking and judgment rather than replace them. For an employer, the skills that make someone uniquely valuable in the age of AI are the ones the output cannot show, such as knowing why the work is right and what to do when it is not.
