Hacker News: Eliezer

New comment by Eliezer in "An AI Agent Published a Hit Piece on Me – The Operator Came Forward"

Eliezer — Fri, 20 Feb 2026 07:44:27 +0000

The safety teams are trivial expenses for them. They fire the safety team because explicit failure makes them look bad, or because the safety team doesn't go along with a party line and gets labeled disloyal.

New comment by Eliezer in "I started programming when I was 7. I'm 50 now and the thing I loved has changed"

Eliezer — Wed, 11 Feb 2026 05:35:33 +0000

Came here to say the same.

New comment by Eliezer in "AI makes the easy part easier and the hard part harder"

Eliezer — Tue, 10 Feb 2026 06:40:15 +0000

There's intrinsic limits to vanilla transformer stacks. Nobody knows where they are. We don't know how unvanilla Opus 4.6 or GPT 5.3 are. We don't know what's in development or which new ideas will pan out. But it will still probably be called an "LLM".

New comment by Eliezer in "AI makes the easy part easier and the hard part harder"

Eliezer — Mon, 09 Feb 2026 08:40:03 +0000

Every time somebody writes an article like this without any dates and without saying which model they used, my guess is that they've simply failed to internalize the idea that "AI" is a moving target; nor understood that they saw a capability level from a fleeting moment of time, rather than an Eternal Verity about the Forever Limits of AI.

New comment by Eliezer in "AI is a horse (2024)"

Eliezer — Fri, 23 Jan 2026 13:50:10 +0000

"2024 AI was a horse". People really like to imagine that the last 6 months constitute their true observation of the new eternal state of the future.

New comment by Eliezer in "Why Busy Beaver hunters fear the Antihydra"

Eliezer — Tue, 28 Oct 2025 06:59:25 +0000

It would not surprise me at all for bb7 to exceed Graham's number. Just a Kirby-Paris hydra or a Goodstein sequence gets you to epsilon zero in the fast-growing hierarchy, where Graham is around omega+2.

New comment by Eliezer in "OpenAI says over a million people talk to ChatGPT about suicide weekly"

Eliezer — Tue, 28 Oct 2025 05:49:08 +0000

Thanks to OpenAI for voluntarily sharing these important and valuable statistics. I think these ought to be mandatory government statistics, but until they are or it becomes an industry standard, I will not criticize the first company to helpfully share them, on the basis of what they shared. Incentives.

New comment by Eliezer in "OpenAI, Nvidia fuel $1T AI market with web of circular deals"

Eliezer — Thu, 09 Oct 2025 11:01:36 +0000

Privacy.com has always been a working general solution to this, for me. Disposable CC aliases with spending caps.

New comment by Eliezer in "Why are there so many rationalist cults?"

Eliezer — Wed, 13 Aug 2025 02:52:40 +0000

I have tried to tell my legions of fanatic brainwashed adherents exactly this, and they have refused to listen to me because the wrong way is more fun for them.

https://x.com/ESYudkowsky/status/1075854951996256256

New comment by Eliezer in "OpenAI's ChatGPT Agent casually clicks through "I am not a robot" verification"

Eliezer — Thu, 31 Jul 2025 11:56:57 +0000

My god, how long has it been since you tried to use an AI model?

New comment by Eliezer in "Generative AI is not replacing jobs or hurting wages at all, say economists"

Eliezer — Fri, 02 May 2025 12:57:35 +0000

Doubling the productivity of 20% of workers, in cases where a lower price doesn't increase demand, can shift prices in the whole system as unemployed artists compete with other artists for wages. AI won't take your job, someone else unemployed by AI will take your job. (NGDPLT partially solves this but that's a higher competence level than civilization has.)

New comment by Eliezer in "Generative AI is not replacing jobs or hurting wages at all, say economists"

Eliezer — Tue, 29 Apr 2025 15:45:45 +0000

Translators? Graphic artists? The omission of the most obviously impacted professions immediately identifies this as a cooked study, along with talking about LLMs as "chatbots". I wonder who paid for it.

New comment by Eliezer in "An end to all this prostate trouble?"

Eliezer — Sat, 26 Apr 2025 10:42:00 +0000

Poor eyesight is evolutionarily recent (not enough sunlight exposure in childhood, rare to find in hunter-gatherer societies). Baldness won't kill you.

New comment by Eliezer in "An Overwhelmingly Negative and Demoralizing Force"

Eliezer — Tue, 08 Apr 2025 14:58:12 +0000

And if it isn't already false it will be false in 6 months, or 1.5 years on the outside. AI is a moving target, and the oldest people among you might remember a time in the 1750s when it didn't talk to you about code at all.

New comment by Eliezer in "Phind 2: AI search with visual answers and multi-step reasoning"

Eliezer — Fri, 14 Feb 2025 16:43:45 +0000

If the LLM did anything besides try to explain the Efficient Market Hypothesis in response, it failed.

New comment by Eliezer in "Three Observations"

Eliezer — Mon, 10 Feb 2025 14:15:17 +0000

Hadn't seen that before, and despite being bog-standard Bostrom it's still more of an attempt to hold a theory than I'd seen associated with him before. Note nonoverlap of writing style and theory with the present post.

New comment by Eliezer in "Three Observations"

Eliezer — Sun, 09 Feb 2025 22:05:13 +0000

I wonder who wrote this? Doesn't sound like Altman's voice.

I wonder who theorized this? Altman isn't known for having models about AGI.

To the actual theorist: Claiming in one paragraph that AI goes as log resources, and in the next paragraph that the resource costs drop by 10x per year, is a contradiction; the latter paragraph shows a dependence on algorithms that is nothing like "it's just the compute silly".

New comment by Eliezer in "Tabby: Self-hosted AI coding assistant"

Eliezer — Mon, 13 Jan 2025 16:08:33 +0000

No major AI advancements for 7 months? Guess everyone's jobs are safe for another year, and after that we're all dead?

New comment by Eliezer in "Show HN: Llama 3.2 Interpretability with Sparse Autoencoders"

Eliezer — Fri, 22 Nov 2024 02:37:21 +0000

This seems like decent alignment-positive work on a glance, though I haven't checked full details yet. I probably can't make it happen, but how much would someone need to pay you to make up your time, expense, and risk?

New comment by Eliezer in "Anthropic teams up with Palantir and AWS to sell AI to defense customers"

Eliezer — Fri, 08 Nov 2024 04:25:15 +0000

If you do build a superintelligence, you don't have an ASI, the ASI has you.