Terence Tao: creative strategies, this aspect of LLM tools is still weak(mathstodon.xyz)
mathstodon.xyz
Terence Tao: creative strategies, this aspect of LLM tools is still weak
https://mathstodon.xyz/@tao/113142753409304792
7 comments
https://news.ycombinator.com/item?id=41540902 ("Terence Tao on O1 (mathstodon.xyz)", 432 comments)
Comments moved thither. Thanks!
It's not a copy, it's not a dupe, it's a genuine update.
I think it's an error to think of this as a dupe, and I think it's an error to move the comments there.
But too late now.
I think it's an error to think of this as a dupe, and I think it's an error to move the comments there.
But too late now.
It's not too late. But the criterion here isn't whether the articles are technically different, it's whether the discussions are substantially the same or not. In this case the answer was yes: the discussion in the current thread was generic and therefore indistinguishable from the discussion in the previous thread. You can see this for yourself—here are the subthreads I moved:
https://news.ycombinator.com/item?id=41549800
https://news.ycombinator.com/item?id=41549690
https://news.ycombinator.com/item?id=41549645
https://news.ycombinator.com/item?id=41549625
https://news.ycombinator.com/item?id=41549535
https://news.ycombinator.com/item?id=41549483
https://news.ycombinator.com/item?id=41549342
https://news.ycombinator.com/item?id=41549200
https://news.ycombinator.com/item?id=41549800
https://news.ycombinator.com/item?id=41549690
https://news.ycombinator.com/item?id=41549645
https://news.ycombinator.com/item?id=41549625
https://news.ycombinator.com/item?id=41549535
https://news.ycombinator.com/item?id=41549483
https://news.ycombinator.com/item?id=41549342
https://news.ycombinator.com/item?id=41549200
Related:
Terence Tao on O1
https://news.ycombinator.com/item?id=41540902
Terence Tao on O1
https://news.ycombinator.com/item?id=41540902
Terence Tao has a remarkable ability (among many others obviously) to convey information that I think a lot of us may intuitively understand but struggle to explain clearly.
> Here the result was mildly disappointing: Essentially the model proposed the same strategy that was already identified in the most recent work on the problem (and which I restated in the blog post
This was my concern when the first post hit HN.
Tao seemed excited that “this time around” the “new” model proposed Cramer’s Theorem:
> This time around, Cramer's theorem was identified and a perfectly satisfactory answer was given.
All I could think was, well of course it did now, it has likely since been trained on Tao’s blog posts talking about its failure to suggest Cramers Theorem.
I’m unsure why Tao is allowing himself to be an advertisement for this company and its paid product.
This was my concern when the first post hit HN.
Tao seemed excited that “this time around” the “new” model proposed Cramer’s Theorem:
> This time around, Cramer's theorem was identified and a perfectly satisfactory answer was given.
All I could think was, well of course it did now, it has likely since been trained on Tao’s blog posts talking about its failure to suggest Cramers Theorem.
I’m unsure why Tao is allowing himself to be an advertisement for this company and its paid product.