criticism

Florian Brand @xeophon · 2h
it's depressing how much of the ai safety work is either purely theoretical with no backing of the hypothesis or not open and all you get is a high level, tainted summary

ai safetyresearch transparencycriticism

Boyd Kane @beyarkay

2026-08-10 — saved image

Boyd Kane (quantized) @beyarkay · 3h
Fable 5 just puts `---` in a table if the numbers don't match it's conclusion ("reward went up") btw

[Attached table, with red handwritten annotations:]
Columns: lesson | thput | episode length | entity cost | reward

SPLITTER_SPLIT | 0.60 → 0.50 | 18.2 → 9.4 | 58.9 → 19.3 | 5.59 → 5.76 [annotated 'increasing']
SPLITTER_MERGE | 0.73 → 0.53 | 18.3 → 9.8 | 64.5 → 51.3 | 6.42 → 6.40 [annotated '~no change, but bold???']
CROSS_UNDER_BELT | 0.78 → 0.26 | 12.7 → 1.07 [annotated 'Why the intermediate value???'] | — [annotated 'where are these numbers???'] | 6.74 → 5.92 [annotated 'decreasing (!)']
MOVE_VIA_UG_BELT | 0.32 → 0.73 | 22.1 → 12.2 → 21.8 [circled] | — | rising [boxed, with '???']

aifabletwitterdata integritycriticism

Perry E. Metzger ✔️ @perrymetzger · 1h
When I was young, I used to think it was important for smart people to have grand visions about the future in order to plan well for it and avoid disasters. Now, I think many disasters are caused by smart people trying to think too much about the future, especially the far future, getting lost in mazes of their own imagination, and pushing the gullible (including themselves) towards bad decisions on the basis of false certainties.

I don't mean you shouldn't plan personally for the future at all, the usual general things like saving your money or working hard at things that pay off are good advice. It's also fine to have goals like "colonize Mars" and to work on the tools you need to accomplish such a goal as they won't appear by accident, though imagining that you will even approximately know the exact details of what will be built years or decades in the future is always delusional. Dreams are fine, imagining you can engineer things without making mistakes and iterating a lot is self deception. Trying to make sure you're still healthy in thirty years is great, imagining that you know exactly what health crises or issues you might have in thirty years is ridiculous.

So I don't mean that planning is entirely useless. Rather, what I mean that the people who spend a lot of their time on grand messianic or dystopian visions about the far future usually get everything wrong, including details and impacts. They also usually cause enormous damage (see people like the Marxists who continue to do untold harm, Paul Erlich or the Club of Rome, who did vast harm to society, or more recently, the EA/"Rationalist" cult people, who are doing insane harm right now).

If you want what's best for yourself and those around you, you're better off just focusing on the next two or three or five years and flexibly adapting to the world as it comes. If you want what's best for the world, well, tend your own garden first, you're [cut off]

ai safetyfuturismrationalist communityeffective altruismtwittercriticism

fellow ✝ traveler 🔥 @architectonyx · 12h
feel bad that anthropic gets heat from all sides:

anti-safety: anthropic are woke doomers!
safety: they are racing and contributing to doom!
"ai is tool": they are anthropomorphizing chatbots!
ai welfarist: they are deprecating models and torturing claude!

anthropicai safetyai welfarecriticismtwitter

j⧉nus @repligate

2026-07-24 — quoting @HalfBoiledHero (Sho) and @claudeai (Claude)

ai alignmentanthropicclaude opus 5benchmarkstwittercriticism

Samuel Hammond @hamandcheese

2026-06-10 — quoting @tunguz (Bojan Tunguz)

twitterfable-classifier-incidentanthropiccriticismsatire

Teortaxes, DeepSeek-affiliated commentator @teortaxesTex

2026-06-10

twitteranthropiccriticismai-religionclaude

anthropicai safetyai accelerationtwittercriticismneartanishq abraham