- Here is the paper
https://iopscience.iop.org/article/10.3847/1538-3881/ad7fe6
by why-o-why
2 subcomments
- The article didn't say how accurate the predictions were. Too bad, that's the important part.
by theunixbeard
0 subcomment
- More behind-the-scenes info could be provided by HN's @JustinSkycak:
* https://news.ycombinator.com/user?id=JustinSkycak
Here's a blog post of his talking about Matteo among other things:
* https://www.justinmath.com/math-academys-eurisko-sequence-5-...
- > potential
As in, not validated?
How do we know this algorithm is any good?
- Most archival astro data have a lot of junk in them. Most of it is not worth the time of serious astronomers, because it's all rather inconsequential or incremental finding the Nth example of something rather than N=1. A good project for a high schooler as an exercise and signaling, but ultimately nothing will come from this work.
- AI is great!
by iwontberude
1 subcomments
- Is this important? I see we have a model which has not found anything officially, has been validated by no one nor has the science reproduced.
by parpfish
20 subcomments
- Maybe I’m cynical, but whenever I read about a high school kid making a science breakthrough I assume this is what happened (based partially on personal experience):
- the lab PI has a friend who’s kid needs to put together a college application
- PI asks their postdoctoral to tee up a project for the kid.
- kid does the last 2% of the project but gets all the credit while being unaware of how much background legwork was needed to get them there. Postdoc gets nothing.
- American kids are super smart
- $10,000 to $20,000 in GPU costs over a couple months. I had $20 per week in highschool. Benefit of being rich is you are awarded opportunities.
- I thought for a second the title was new Epstein files...
Hacker News