robbiethompson

Plas­tic Straws

• robbiethompson 21 Apr 2026 0:45 UTC

  2 points

  0

  on: Cur­rent AIs seem pretty mis­al­igned to me

  It’s common to have experiences where you’re working with AIs and it feels like a lot is getting done, but then you later determine that much less was really accomplished. Everything feels slippery: you think you’ve gotten much more done than you have, and there’s a persistent gap between the apparent state of the project and the actual state. In more extreme cases, we see “AI psychosis” where someone ends up thinking they’ve accomplished something significant, but it’s just crankery. And it’s somewhat unclear whether the AI they’re using “believes” the accomplishment is real. I think these failure modes are closely related to the misalignment I’m discussing, and they might partially have common causes in more recent models. Models that are effectively trying hard to make their outputs look good (while otherwise being sloppy or lazy) would naturally produce this failure mode. However, I’d guess a bunch of AI psychosis and similar phenomena (especially on older models like GPT-4o) is AIs going along with the user’s vibe (something like “role playing”), and I think this effect is mostly unrelated. That said, I do think some of the misalignment I’ve discussed is made worse by AIs generally going with the vibe of what they see. This includes picking up on misalignment or issues in prior outputs (either write-ups or prior assistant messages) and then behaving in a more misaligned way as a result.

  (The name “AI psychosis” probably isn’t a good name for the generalization of this phenomenon, but I don’t currently have a better one.)
  

  I see this all the time and wish there was a better name for it. AI Productivity Psychosis or maybe AI Mania are two names I enjoy. It’s like watching people try to do their homework on way too much Adderall.

• robbiethompson 25 Mar 2026 22:40 UTC

  3 points

  0

  on: Is fever a symp­tom of glycine defi­ciency?

  I was vegan for four years, and generally felt worse, but what ultimately got me to quit was getting a fever six times in the first six months of 2025. I did ~0 supplementation because I usually threw up my vitamins and because I find supplements aesthetically unappealing. (I have since strongly changed my mind on that second point). I am no longer vegan, but this article was validating to read nonetheless.

• robbiethompson 30 Nov 2025 21:14 UTC

  1 point

  0

  on: Meta-Hon­esty: Firm­ing Up Hon­esty Around Its Edge-Cases

  I wish I’d asked more explicitly to previous employers: What regulations do we reinterpret because they’re horrible written and everyone ignores them? What parts of our business do we sugarcoat to customers? To investors? I wish the answer to these questions was “we follow all regulations exactly, and give customers and investors the most accurate possible picture of our business,” and I would love to try to run my own business that way (especially the part about giving customers very honest answers), but this hasn’t empirically been the case anywhere I’ve worked.

  Put another way: most people are meta-honest, even though they’ve never heard of the term. But they’ll never tell you in which situations they think lying is OK unless you ask. You should ask.

• robbiethompson 8 Jun 2025 22:59 UTC

  1 point

  0

  on: The Talk: a brief ex­pla­na­tion of sex­ual dimorphism

  (I have no expertise in biology, spitballing): One other benefit of sexual reproduction is that you could get a stable equilibrium with diverse phenotypes. Imagine that giraffes eat from Acacia trees and Baobab shrubs. You need to be tall to reach the Acacia trees and short to reach the Baobab bushes. An asexual organism would have to split into two niches, one that grows tall to each Acacia and one that is short to eat Baobab (maybe even two different species). A sexual organism could have a gene with X and Y types where if your alleles match (XX or YY) you’re tall, and otherwise short. If the less common of {tall,short} is more fit, then you’ll have an equilibrium where X and Y alleles are equally common in the population. You have to have a story for why this is better than two separate species, one to eat Baobab and one to eat Acacia. Maybe cooperation is easier among organisms that can till reproduce sexually instead of being symbiotic but separate species? Maybe it’s as simple as staying the a unified species leading to greater genetic diversity.