3 ms·
this gets more nuanced because "the sigmoids won't save you": https://www.astralcodexten.com/p/the-sigmoids-wont-save-you https://www.astralcodexten.com/p/the-s
by bryan0 2mo ago
this gets more nuanced because "the sigmoids won't save you": https://www.astralcodexten.com/p/the-sigmoids-wont-save-you https://www.astralcodexten.com/p/the-sigmoids-wont-save-you
- throwaway27448 2mo agoIf the sigmoid is incorrect it's certainly more correct than the exponential. > https://www.astralcodexten.com/p/the-sigmoids-wont-save-you https://www.astralcodexten.com/p/the-sigmoids-wont-save-you The conclusion of this article seems to be "you should give ai the benefit of the doubt against all reason". Barf
- conformist 2mo agoIsn’t the point more “it’s easy to fall into the trap to believe that predicting when the sigmoid is going to bend is possible and the right heuristic is to instead extrapolate locally”? That aside, I’d question whether applying the Lindy effect in particular to something that’s not really a life expectancy but more a growth rate is credible… or perhaps a bit circular since it “assumes away” the ceiling.
- throwaway27448 2mo agoNobody in this thread is trying to predict when the sigmoid is going to bend. Perhaps they should
- yazaddaruvala 2mo agoPredicting when the sigmoid bends is difficult and predicting how long until it unbends is equally difficult. The simpler assumption is that over enough time, the S functions stack together for long enough that working backwards from exponential is a better predictor of reality. These stacked S curves have continually been true with most technology.
- haldujai 2mo agoIt hasn’t bent already? 2022-2024 certainly seemed far more exponential than 2024 to present.
- NiloCK 2mo agoI find this astounding. 2024 to present thread is can write a coherent 15 line function to ... what exactly? No future for research mathematicians othet than as tastemakers / agenda setters?
- aurareturn 2mo agoA lot of HN posters thought it was already bending at GPT4o/GPT4.5. Turns out, it kept accelerating (at agentic tasks).
- throwaway27448 2mo agoI suppose it depends what you're expecting. From my perspective it bent around november 2025
- annzabelle 2mo agoThe author of that post is a prominent Bay Area "rationalist," who have had a quasi-theistic relationship with the concept of all-powerful AIs for a couple decades now.
- mitthrowaway2 2mo agoSo are they right or wrong about the sigmoids?
- annzabelle 2mo agoI dunno. They've been saying this for decades, including in a very well read Harry Potter fanfic, and I'd always dismissed them as kooks, but maybe they're right in the end.
- synarchefriend 2mo agoYou are confusing Scott Alexander with Eliezer Yudkowsky.
- annzabelle 2mo agoScott Alexander is a follower of Yudkowski and comes out of that general LessWrong space. This is the plural they, not the singular they.
- mitthrowaway2 2mo agoSaying the same thing for decades, when things progress over those decades along the general trend line that you were worried it would, is decent evidence of a prescient prediction. Climate scientists also seemed pretty kooky in the 80s when they sketched out their trendlines, but now their concerns are all over the headlines as reality caught up, and the same is becoming true of the AI safety people.
- vasco 2mo agoAll of the point of that article is that most people that think something is a sigmoid think it'll bend just as they are publishing their analysis. And the article says, don't do that, assume it'll be related to how long we've been on the "goes up" part. Nothing in that article says it's not a sigmoid.