[freetonik]

The idea is great, and accessible explanations of cutting edge science research is definitely something the society can benefit from. Especially to combat clickbait-y publications of some journalists who transform scientific papers into basically false claims.

But to be honest, AI-based summaries seem like a dangerous way of handling these. The danger is that the summaries, for the most part, would probably be accurate enough and well "written" so that the reader would feel confident trusting it, but it's possible for AI to come up with a completely misleading or false claim, and it'll be well hidden in the "good" content of the summary.

And then it'll take humans to check and re-check the text anyway, which feels very similar to how AI is used in software development: at some point it's counter-productive because you have to check the generated solution instead of generating it yourself.

[Bjartr]

That doesn't sound all that difference from pop science writing done without AI.