If it were submitted to a journal, it would be outright rejected. We should give the peer review process -however flawed it is- some credit here. Dumping hundreds of AI-generated preprint articles on the field should not be allowed.
No. Letting some tiny handful of individuals selected by a tiny set of review-delegators review this is immeasurably worse than just releasing it publicly and letting anyone with the sufficient expertise review it. You know nothing of how peer review actually works, or have not thought about why that process would be only harmful in this case.
Normal peer review works by submitting the article to the appropriate journal, suggesting some preferred reviewers that have expertise in the subject, after which the rest of the process ideally is then orchestrated by an experienced editor.
If a journal receives a paper that is unreadable, it is outright rejected with the comment that it should be made readable, regardless the results in that paper. What is the point of having results if you are unable to convince your audience of those results? You could just as well just sit on them and never communicate them. What is harmful about this process?
Try thinking for just two seconds about what your process would result in, given the scale and amount of results here. You are basically asking for these results to become available to all experts only months / years from now, only after some select experts are chosen and first get to read these results. If there is any error or bias in that selection process, the whole thing is delayed even further or lost to the file drawer entirely. Or, since you say these are all unreadable, and should be rejected, you are in fact arguing these results should never be released at all.
Which is why it is obvious post-publication peer review is the only sane way to handle this. Formal peer review is not even remotely up to the task here.
This makes me think: how would someone like Mark Zuckerberg handle this for his own internet presence? Or does he sidestep this issue completely by having assistants for nearly everything? I can imagine he doesn't do much more than look at .ppts and fire off emails. Do data brokers have any information at all on this guy?
Not just lawyers. You get a personal physical and cybersecurity team in most cases. Most largeish businesses also have a dedicated physical security team as well.
You as a normal individual simply cannot replicate the kinds of services that an Executive Protection services provide [0][1].
It would be funny, if there was a website anywhere, which accumulates data about all the tech giant C levels and shares it with the public, just like they share data about all of use behind closed doors to manipulate us and sell us shit, or sell our data to the next tier of data hungry businesses.
Probably wouldn't last long though, as they would be furious, that us lowly human beings are able to glean anything about them. The double standard of this is not obvious to them.
I actually convert the SDFs to voxels and render triangles, but.. some people have written direct SDF renderers. There's a guy that posted to YouTube not too long ago that wrote a really cool one with physics and everything
The weird thing is I've seen LLMs "typo" stuff pretty often. Yesterday I asked Gemini a question about the Python Twisted framework and it answered about Deferreds but misspelled it as "Deferends" in one spot.
Pi * r2 (squared) both pans. Divide smaller pan area by larger pan, now you have the % of how much the smaller pan recipe fills up the larger pan, and the missing % you need to fill. Increase ingredients by that % divided by the filled %.
Small pan area: 20 sq cm
Larger pan: 48 sq cm
20 / 48 = .42, I'm missing .58 of the pan. .58 / .42 is 1.38. My recipe needs 2.38x the original to fill the larger pie pan.
Having used Claude a lot for the past four months, I get exhausted from its style. Every sentence has been buffed and polished and enhanced and it never just plainly says a thing. It's as if you have to eat every meal with every possible flavor enhancer in existence. At one point I will totally lose it if it mentions "smoke test" one more time.
It is also totally not what I want for docstrings in code or Powerpoint drafts, where consistency in language is critical.
Like I said in another thread, everything Claude (Code) writes sounds like it’s congratulating itself of how smart it is. While not actually being that smart.
I believe AI technical prose peaked with Opus 4.6. I still use it (mostly for that purpose) and I think it's legitimately great. I'm hoping they are able to reverse the trends that benchmaxxing and RLAIF have wrought on Claude's non-code output with the future models.
Now that the public anger is growing i really hope Anthropics next models seriously address this. Or for that matter, i hope everyone (OpenAI/etc) starts focusing on this too.
Intelligence is important but it feels as if the LLM is only capable of conveying information while drenching it in vomit.
Claude Code's creative use of language in an engineering context is often quite infuriating. I want language to be as simple as possible, it should be accessible and require as little context as possible. Below are a few examples that Code wrote that I encountered in my own work (related to signal processing):
"excursion"
(means: a spike/jump — a value that rises or deviates from baseline, just say spike or outlier)
"legitimate majority-normal baseline"
(means: a real majority of normal pixels)
"matched pool of pure-noise ('normal') pixels"
(means: the same number of normal pixels, don't bring pools into this)
"ablation"
(means: comparing before vs. after — turning a thing on/off to see what changes)
I actually used Claude Code to try and find examples like these but not entirely unexpectedly it had a very hard time detecting these, even though I encounter them like every other sentence. I can imagine this style is great when writing prose, not so much when writing documentation.
Similarly, I tried using Claude Design to make some slides, and the individual slides were fine, but just about every other slide it used different words to denote the same concept, which I would say is the fastest way to lose your audience.
