Hacker News: singularity2001

New comment by singularity2001 in "GLM-5.3-Flash"

singularity2001 — Thu, 27 Aug 2026 10:23:51 +0000

Yes, definitely blocked.

New comment by singularity2001 in "Humanity has the debate about AI consciousness backwards"

singularity2001 — Thu, 27 Aug 2026 10:23:12 +0000

Consciousness is not binary and has many aspects which might be valued from 0 to 10.

New comment by singularity2001 in "GLM-5.3-Flash"

singularity2001 — Wed, 26 Aug 2026 16:02:41 +0000

At the current 50%-off GLM-5.3-Flash price ($0.075/M input, $0.25/M output; cached input $0.015/M), surprisingly, roughly $400–900/month would buy token throughput comparable to fully exhausting Claude Max 20×

New comment by singularity2001 in "GLM-5.3-Flash"

singularity2001 — Wed, 26 Aug 2026 15:52:47 +0000

I blocked Z.ai as soon as they were loading 10 different external providers including Alibaba who was just proven to execute silent sound fingerprinting mechanisms.

New comment by singularity2001 in "One corner of China’s internet is insisting that the Tang Dynasty never existed"

singularity2001 — Tue, 25 Aug 2026 10:28:41 +0000

Some also claim that bronze metallurgy wasn't brought together with the horse, sheep, cattle and chariot from the west.

New comment by singularity2001 in "Don't Paste the AI, please"

singularity2001 — Thu, 20 Aug 2026 09:27:49 +0000

Worst case you just ignore it. Best case there is something useful in it. Shrug.

New comment by singularity2001 in "Claude: System Prompts"

singularity2001 — Sun, 16 Aug 2026 19:31:09 +0000

Also, why don't they bake in these limitations via reinforcement learning so they can keep the prompt context clear.

New comment by singularity2001 in "AI has access to a vastly larger working memory than the human brain"

singularity2001 — Sun, 16 Aug 2026 08:40:55 +0000

it's a paradox given how extremely complex each single neutron is. my current solution is that most of the complexity of neurons is just for maintenance

New comment by singularity2001 in "The Indo-European Family Tree"

singularity2001 — Fri, 14 Aug 2026 06:18:16 +0000

4000 BC? I don't buy it: if you look at how little Greek changed over the last 3500 years and compared to how different it is to other languages this doesn't sit right

New comment by singularity2001 in "OpenAI's new device will be hockey puck-sized and cost over $300"

singularity2001 — Wed, 12 Aug 2026 05:34:16 +0000

in the US you may ask Siri about stuff in Europe it still takes about half a minute to activate GPT and talk to it. Investing $1000 in a new iPhone might solve the problem soon but at this point I'd rather try a new $300 device just out of spite

New comment by singularity2001 in "Artificial Intelligence used to design new viruses"

singularity2001 — Sun, 09 Aug 2026 06:45:01 +0000

I don't want it to be a lab leak but all the evidence points towards it being a lab leak

New comment by singularity2001 in "How real are real numbers? (2004)"

singularity2001 — Tue, 28 Jul 2026 18:58:21 +0000

there is a practical reason to get rid of the fantasy reals and restrict oneself to normal reals or some other new invention:

Since Lean has become more popular as a proving system I've stumbled upon one very annoying feature of reals: they are not computably comparable. The system says you can never know whether two arbitrary real numbers are the same because you don't have enough time to compare them.

New comment by singularity2001 in "How real are real numbers? (2004)"

singularity2001 — Tue, 28 Jul 2026 18:51:34 +0000

what he probably meant is that if you write down pi with 200 digits there's nowhere in the universe where you can find pi with that precision

New comment by singularity2001 in "Ask HN: What are the most promising RL fields for a new master student?"

singularity2001 — Mon, 27 Jul 2026 10:40:18 +0000

World Models, really big world models like Dreamer V4, which unfortunately you can't do easily as a master student, but maybe you can work with tiny versions and prove that the direction works, or Lora some foundation model. If you look at ARC 3, it is clear that the reasoning capabilities of language models need to be combined with 'reflexes' of classic reinforcement learning.

New comment by singularity2001 in "We have proof automation now"

singularity2001 — Mon, 27 Jul 2026 10:28:03 +0000

Have you tried writing any programs in it?

New comment by singularity2001 in "Terrence Tao's ChatGPT Conversation about the Jacobian Conjecture Counterexample"

singularity2001 — Wed, 22 Jul 2026 22:23:13 +0000

Maybe Terrence is chasing too many red herrings. Maybe he should just have asked it to find a counter example to the conjecture and on success explain how it was found.

New comment by singularity2001 in "Terrence Tao's ChatGPT Conversation about the Jacobian Conjecture Counterexample"

singularity2001 — Wed, 22 Jul 2026 21:51:08 +0000

You are badly informed the counter-example was found shortly before. Terrence just tried to replicate how it was found.

New comment by singularity2001 in "Terence Tao's ChatGPT conversation about the Jacobian Conjecture counterexample"

singularity2001 — Wed, 22 Jul 2026 21:37:08 +0000

they're still bounded by training data and how it relates to the real world.

Yes and no. They can extrapolate and build upon the training data, as was the case with the last dozens of math proofs

New comment by singularity2001 in "Terence Tao's ChatGPT conversation about the Jacobian Conjecture counterexample"

singularity2001 — Wed, 22 Jul 2026 21:35:13 +0000

They will have decades of experience with AI and they will be able to guide them by sniffing their hallucinations from single words.

New comment by singularity2001 in "Linux 0.11 rewritten in idiomatic Rust, boots in QEMU"

singularity2001 — Tue, 14 Jul 2026 10:22:20 +0000

The Redox operating system is a complete rewrite of operating system concepts in Rust.

Disclaimer, I worked on the Mac OS port.