Astral Codex Ten 07月14日 14:55
Open Thread 390
index_new5.html
../../../zaker_core/zaker_tpl_static/wap/tpl_guoji1.html

 

本文是一篇开放讨论的周报,涵盖了多个话题,包括对Stephen Skolnick关于精神分裂症微生物的回应,以及对之前的双胞胎同病率模拟的修正。文章还讨论了Gary Marcus对AI赌注的质疑,以及作者对赌注中行为不当的道歉。此外,文章还提到了对Davide Piffer关于种族与智商研究的评论,以及关于抗蛀牙益生菌用户视力下降的案例。最后,文章预告了关于阿尔茨海默症的讨论和Lighthaven活动。

🧠作者对Stephen Skolnick的回应进行了评论,并指出了之前的模拟精神分裂症双胞胎同病率计算中的错误,但认为该模拟仍具有说明意义。

🤖作者回应了Gary Marcus对AI赌注的质疑,认为Marcus的反对意见已在之前的文章中得到解答,并承认在赌注中存在不当行为,为此道歉。

🧐文章讨论了Sasha Gusev对Davide Piffer关于种族与智商研究的评论,并提及了关于抗蛀牙益生菌用户视力下降的案例,作者对此表示谨慎关注。

💡文章还提到了关于阿尔茨海默症的讨论,并鼓励读者参与关于淀粉样蛋白假说的讨论,并预告了Lighthaven活动。

This is the weekly visible open thread. Post about anything you want, ask random questions, whatever. ACX has an unofficial subreddit, Discord, and bulletin board, and in-person meetups around the world. Most content is free, some is subscriber only; you can subscribe here. Also:

1: A few weeks ago I wrote Contra Skolnick On Schizophrenia Microbes. Stephen Skolnick has now responded here. I don’t find his response convincing, for reasons I explain in this comment.

2: But some people pointed out an error in the original post. I said my post a few years ago simulating twin concordance rates in schizophrenia got the exact right answer. After a correction for a math error, it got an answer which - while still sufficing to explain why you shouldn’t expect twin concordance to necessarily equal heritability - was not exactly right. I explain why I think the simulation is illustrative anyway here. You can see a better simulation that does get the right answer here. I’ve edited the Skolnick post to reflect this.

3: Gary Marcus accuses my recent Now I Really Won That AI Bet of being a “straw man” that doesn’t fully engage with the arguments against the existing AI paradigm being unable to master compositionality. He makes his case here. I believe all of Marcus’ objections were already addressed in the original post (CTRL+F “still one discordant note in this story”), except his claim that previous use of these prompts might cause “data contamination” - it’s trivial to demonstrate that 4o succeeds on other prompts of approximately the same difficulty; you can see the comment here for an example.

4: Vitor, the counterparty in my AI bet, concedes but notes I have behaved badly: I slightly shifted terms to deal with model limitations, posted the original (premature) victory claim, retracted the original victory claim, apologized for the original victory claim, and posted the new on-time victory claim, all without consulting him. He’s right that this is bad behavior, and I let my excitement around figuring things out about AI get ahead of the fact that this was a real bet against a real person who deserved to be consulted on how I talked about it. I apologize.

5: My recent Highlights From The Comments On Missing Heritability included a comment by Sasha Gusev criticizing Davide Piffer’s work on race and IQ, and I partly endorsed Gusev’s criticism. Piffer responds here.

6: A user of the anti-cavity tooth probiotic I profiled in 2023 claims to be gradually losing his vision, and proposes a theory that the tooth probiotic’s formate metabolism might be to blame. Commenters mostly seem skeptical (1, 2, 3, 4) citing both theory (it seems like there should be too little formate to matter) and evidence (out of ~1000 users, nobody else has mentioned these symptoms yet); they propose that out of a thousand users, it’s not surprising if one develops a weird disease for unrelated reasons. Still, I am broadcasting this out of an excess of caution. If you are a formate metabolism expert, consider taking a look and weighing it; if you are a probiotic user with similar symptoms, consider speaking up (I don’t really want to be the permament clearinghouse for this, but if you have no other way to speak up, you can email me). To avoid psychosomatic panics, if your “similar symptoms” are along the lines of “now that you mention it, my eyes do maybe feel a little funny”, please wait a few days before freaking out.

7: Some good comments on the Alzheimers mouse review post. I asked if anyone was willing to defend the amyloid hypothesis (I only ever see people attacking it!) and several knowledgeable people weighed in. I might publish something on this later. And if you’re looking for more writing on Alzheimers pathology, Nintil has a series on the topic.

8: Another Lighthaven event ad:

Metagame 2025, a game and puzzle design conference that is also itself an immersive megagame, is happening September 12-14 at Lighthaven in Berkeley. Early bird tickets are $500 through July 17th. If you're interested in getting involved, contact Ricki Heicklen or join the Discord.

Fish AI Reader

Fish AI Reader

AI辅助创作,多种专业模板,深度分析,高质量内容生成。从观点提取到深度思考,FishAI为您提供全方位的创作支持。新版本引入自定义参数,让您的创作更加个性化和精准。

FishAI

FishAI

鱼阅,AI 时代的下一个智能信息助手,助你摆脱信息焦虑

联系邮箱 441953276@qq.com

相关标签

精神分裂症 AI 智力 视力 阿尔茨海默症
相关文章