My art/design students employ AI to generate ideas.
Possibly this project can be used is a similar manner... as a way to start a brainstorming session or suchlike. In my experience of working with research engineers, I see the reluctance to playfully ideate as one of their weakness.
I get why you think that would be vastly more useful it actually is.
Suppose you did the million monkey with typewriter thing and asked not for Shakespeare but any good book. The overwhelming majority of it would be trash but chances are you’re going to find many good books before the exact works of Shakespeare in the order they where written.
Basically the more specific stuff you want the less likely you are to get it. So those million monkeys would type out a decent haiku relatively quickly, any you would get someone’s correctly done tax return more quickly than your tax return.
A math paper that actually solve some problem are much closer to your specific tax return than just anything. Unlike art what’s valid is very constrained so tossing out essentially random stuff and seeing what sticks just isn’t that helpful.
These language models would at least be using random mathematical or scientific ideas so it’s better than the monkey’s but just not by enough to be useful.
Possibly this project can be used is a similar manner... as a way to start a brainstorming session or suchlike. In my experience of working with research engineers, I see the reluctance to playfully ideate as one of their weakness.