I'm glad on not alone on this hill.

And that's the crux of it. It's good work, and contributes, but we cannot extend finding to "education" or institutions. The study isn't conducted in an ecologically valid setting. Which if *fine.*

But let's not make this out to be some sort of amazing evidence against AI. It isn't.

The effect sizes across studies are also modest to tiny. Results are statistically significant sure, but who cares of we're talking about a fractional mean differences in a self report measure on a 1-5 scale? I don't. Not for the broader argument of "is AI good for us/education/educators."

Still, does that extra effort = learning? Maybe? I'm not fully convinced by the author's arguments. They also use a *self-report* measure for learning. Not great, but they use NLP to analyze the written text, which helps.

It's good work. The authors operationalize "effort" based on time on task, which isn't great since it's an apples to oranges comparison. Getting information via ChatGPT is frictionless by design. Google simply works differently. Time differences are predictable. This difference uninteresting.

First, here is the paper this news is based on. Read it so you get a feel for the nuanced findings. You know, do the hard thing we accuse students of not doing.

academic.oup.com/pnasnexus/ar...

Next time submit it in cursive, duh.

Multi-generational homes are thing. Not saying this is that, but I've seen a ton of large homes that are designed for parents + grandparents + maybe siblings and all the kids.