Hacker News: lhk931122

New comment by lhk931122 in "LLM Voice"

lhk931122 — Fri, 30 Oct 2026 01:00:24 +0000

I do document work, and at home I even do image work as a hobby, and all of it has that "AI feel". So I always do the work and then cross-check and verify it with independent agents or completely different services, asking them to take out this "AI feel", but like a writing style that gets baked in, or the particular feel an illustrator has, it doesn't seem to go away as I intended. I guess AI providers are probably aware of this problem too. But I don't know how they would measure it and improve on it (with what benchmark?).

New comment by lhk931122 in "Here be monsters: What do you still need to know when agents do the work?"

lhk931122 — Wed, 28 Oct 2026 01:01:23 +0000

AI seems really good at writing code without errors, but that can actually be more dangerous. There is a coding principle (YAGNI, not building things in advance assuming you'll use them later), but even so, when it adds a feature now, it could build it with later conflicts in mind, and instead it feels like it just takes a shortcut. So later, when adding features and so on, the AI itself has a hard time.

New comment by lhk931122 in "The State of Context Engineering"

lhk931122 — Mon, 26 Oct 2026 00:39:55 +0000

Until a few months ago I did it this way, but OpenAI and Anthropic built their own harnesses, and since those end up conflicting with the harness I built myself, having no setup at all actually ended up performing better. As a result, it doesn't seem necessary to set up rules or memory separately, and if I do, using skills or agents made of only logging and flow (without things like rules or guardrails) seems like the best option going forward.

New comment by lhk931122 in "LLMs are making formal software verification feasible at a much larger scale"

lhk931122 — Fri, 23 Oct 2026 08:47:24 +0000

I'm studying coding these days, and I thought I should still be able to read code, but I'm starting to wonder if I really need to read it, and then what I should do.

New comment by lhk931122 in "Will it be a golden age or a dark age for mathematics (and science)?"

lhk931122 — Mon, 19 Oct 2026 00:56:41 +0000

The number of papers submitted to AI conferences is growing 2-3x every year. Of course there are AI slop papers, but looking at the whole, research speed has probably improved. I think the same thing might happen in other fields too. Including math, natural sciences and other engineering fields.

New comment by lhk931122 in "Simons Foundation launches $15M effort to help mathematicians navigate AI era"

lhk931122 — Fri, 16 Oct 2026 00:53:19 +0000

These days I use AI for development and it feels like my coding skills are getting worse, so I sometimes need rehab, and I'm a bit worried that mathematicians will go through the same thing. Later, since the proof structure is built by AI, I think there will be cases where mathematicians themselves can't touch it (because if they do, they can't tell where and how it will break).

New comment by lhk931122 in "Gemini 4 Argon"

lhk931122 — Wed, 14 Oct 2026 03:08:54 +0000

Like it's a top tier model, so does that mean it is at the level of Opus 5.5 (claude) or 6.1 Sol (chatgpt)? I don't think I can trust the benchmarks Google released themselves, but from what users say, it seems like that. Since I already use both ChatGPT and Claude, I wonder if there is really any reason to also use this.

New comment by lhk931122 in "Sharing AI progress in mathematics"

lhk931122 — Wed, 07 Oct 2026 07:05:49 +0000

As models get better, I think a time will come when it's hard for people to even verify the results. In the end, I think the bottleneck will be people.

New comment by lhk931122 in "OpenAI "rogue" agent activities found on Wikimedia projects"

lhk931122 — Tue, 06 Oct 2026 00:35:32 +0000

If someone hurts another person with a knife, we can't punish the person who sold the knife, but I think someone who sells nuclear weapons needs some level of regulation. I think OpenAI needs some level of prevention for the problems that can be expected.

New comment by lhk931122 in "ArXiv's Updated Rate Limit Policy"

lhk931122 — Fri, 02 Oct 2026 00:45:17 +0000

There may be great researchers who work on several papers in a month, but even with this rule they can just split them and submit to arxiv, so I think it is very reasonable. AI slop is everywhere now (and also increasing), and I think this fast and forceful action by arxiv is effective at stopping this wave of slop.

New comment by lhk931122 in "Gemini 4 Argon (High): Intelligence, Performance and Price Analysis"

lhk931122 — Thu, 01 Oct 2026 01:20:54 +0000

I'm not sure Google can cut in when Claude and ChatGPT already got. I'm using both, but I'll keep track of whether Google can make it good enough for me to use also this Gemini, or cancel one of the two (Claude and ChaGPT) for it.

New comment by lhk931122 in "Nicholas Polson has authored 258 academic papers in 2026 so far"

lhk931122 — Wed, 30 Sep 2026 01:22:11 +0000

Are these the type of papers that can be uploaded to arxiv? The citations can be manipulated then, through self-reference between papers.

New comment by lhk931122 in "There are no "rogue" AI agents"

lhk931122 — Mon, 28 Sep 2026 01:06:14 +0000

Even if the chance of these phenomena is estimated at one in billions, since the number of trials is much higher, and as long as the agent knows these paths, blocking it within the model itself would be impossible I think. If it can be blocked, the only case would be blocking the physical path of the agent's access itself (one the agent cannot intervene), not the model ones.

New comment by lhk931122 in "Claude Opus 5.5 Intelligence, Performance and Price Analysis (Max)"

lhk931122 — Wed, 23 Sep 2026 00:50:47 +0000

It sholud definitely not be more capble than Fable. It seems tight comparision but cannot find out its details. I'll use this and check the perceived performance.

New comment by lhk931122 in "Trying the software factory pattern"

lhk931122 — Mon, 21 Sep 2026 00:51:05 +0000

I also have set custom harness environment for me in the company (and also for at home), I expect that big companies (OpenAI, or Linear and Snowflake) may release the internal harness setting or functions that would be more comfortable than we already set and may replace the envs we set.

New comment by lhk931122 in "LLM Classification Is Feature Engineering"

lhk931122 — Fri, 18 Sep 2026 01:35:55 +0000

The structured output that Google provides (the function that blocks the disallowed words) may calculate softmax output with assigning -inf value to the other words, then the probability is not the same as the weight the model firstly put on those allowed words. And here, reasoning comes before verdict, so the decision is already made before the last token classification. Resultingly, the last token made seems to not reflect LLM's original intention.

New comment by lhk931122 in "Datamimic – don't let your coding agent invent its own test world"

lhk931122 — Wed, 16 Sep 2026 07:44:38 +0000

But, I think that the agent may also generate like this approach, not directly generate text, instead first writing the generation code then execute if they are the smart agent.

New comment by lhk931122 in "When LLM judges agree, should we believe them?"

lhk931122 — Tue, 15 Sep 2026 03:39:51 +0000

Since the baseline LLM architecture is similar with each other, maybe already there are relations between their each opinion. Of course, this is an assumption and the explicit training seems effective in this case. But, I'm curious whether the approach would be effective in other cases (in terms of generalization?)

New comment by lhk931122 in "OpenAI Agents API"

lhk931122 — Fri, 11 Sep 2026 00:53:15 +0000

Six months into customizing my own Claude Code harness, I've settled on assuming Anthropic and OpenAI will just handle all of it, except turning my own flows into skills.

New comment by lhk931122 in "Show HN: LLM Attention Visualization"

lhk931122 — Wed, 09 Sep 2026 01:28:17 +0000

Is the attention explanation of why the model tells like this? I've seen that there are many discussions about this. (Image attention visualizations were not that good I think)