When physicist & best selling author Chanda Prescod-Weinstein sent me this declaration on AI written & signed by a bunch of mathematicians, I was initially super skeptical because I had seen a whole bunch of such “declarations” which, at best, are distractions from the actual problems we should be discussing. But once I started reading, I was mostly nodding my head up and down in complete agreement and super impressed at how spot on so many of the points were.
11
Replies
·
0
Boosts
The link doesn't work for me. Too much traffic maybe?
@timnitGebru re: the recent Jacobian conjecture counter-example, I don't have the mathematical training to verify the Lean program. AFAIU, there is still work to be done by professional mathematicians to verify this work, but the author writes like the LLM definitely solved the problem. not to mention the whole article reads like one long promo.
it seems like a distilled model trained specifically for use in mathematics would be much more efficient than a general LLM with trillions of params...
@timnitGebru My mom is a mathematician, long since retired from working as a Systems Analyst. I had asked her what her thoughts were on AI being used for complex mathematical computations and her response was pretty succinct, though I'll remove the curse words as mom is from Brooklyn and curses like a sailor. "When I had to model the stress distribution on an unused hull model over time, I couldn't use any existing model, they didn't account for any of the statistical noise unique to this model. I had to create entirely new Kalman Filters to do it. AI can't create, it can only copy."
Side note, just to kvell a bit about my mom, some of those filtration algorithms she created are still in use today. They were novel enough at the time she created them that JPL patented them and named them after her.
Given another round of hype from these companies, now is as good a day as any to share. Some points from the document:
➡️ “Mathematical arguments are regarded as transparent and subject to independent verification. They may be extremely long or difficult, but in principle no proprietary knowledge or equipment should be required to understand them.”
➡️“Mathematicians share a concern for proper evaluation of mathematical work relative to shared standards of depth, difficulty, and significance.”
➡️“Technologies which affect the way in which mathematics is practiced may disturb the current system of incentives. The use of artificial intelligence — and thus also the sort of problems which it can address — may become incentivized for its own sake, disrupting our mechanisms for hiring, funding, and recognition.”
“Proper evaluation is endangered if results are communicated through informal channels such as press releases or blog posts, often without any research paper or other disclosure of information necessary for scientific evaluation. This practice seeks publicity for new results on market timelines before the accepted processes of community evaluation in mathematics can take place....
In many cases this leads to simplifications in reporting, such as overemphasizing the significance of automated tools and undervaluing the prior human contributions which have made those tools possible. Such oversimplification risks influencing public opinion in a way that not only damages perceptions of mathematics, but also misleadingly uses specific mathematical tasks as metrics for the general reasoning capacities of commercial products.”
