As with most every place else in organizational life, there is pressure on the learning function to use more AI. In higher education, we have the same issue—probably worse because the two main ways students have always learned in college don't work anymore. That's reading and writing.
You probably know that ChatGPT or any of the other LLMs can write perfectly good term papers. In fact, I would argue that they are better at that task than anything else. The reason is that the papers don't have to be great, just pretty good. The topic for most any term paper has been written about before, maybe many times, so there is plenty of material on the internet from which to construct a summary paper.
What you might not know is that they are even good at drafting fake papers about any particular student! An assignment asking them to write about their own experiences still gets written by AI (although we have had students mistakenly submit the prompt they gave ChatGPT instead of the paper it produced … ). The prompt would be detailed like, "describe what a 20-year-old student would experience working for the first time in a New York investment bank."If they want it to be longer, they just add more details about themselves such as "the person is a woman English major who has only had a few business classes."
The point of writing papers is not just or even mainly to assess what a student has learned.We could still do that with in-class essay tests hand-written in blue books—many of us still shudder seeing those. (The only drawback is that students no longer know how to write cursive.) The point is you learn to think by having to assemble a paper, figure out what you really want to say, check which evidence makes sense and so forth. Learning that way is probably gone.
What might be more surprising is that students don't read the assigned materials, either, because the LLMs produce short summaries instead. You may remember "Cliff Notes" summaries of great books? Not quite the same as actually reading Macbeth. But now, imagine that play is condensed by AI to less than one page. Without reading and writing, we have to find other ways to learn.
Here's what I found that really is useful, students like it and it uses AI. We have long had computer-driven simulations about businesses that students can play; they like them, and you can learn from them. But what if the learning we are after is about people rather than businesses, and what we are manipulating is what we say rather than numbers? Role-plays are really useful here, but the drawback is that they are only as good as your partner in the exercise. If the partner can't be an effective actor, it's not going to work well, and if you want to learn something about how to be a supervisor, as in my class, only one person can play that role at a time.
Read the full article here: