Lockal

That is not dumb - erasing copyrights is just their business, and even when caught has zero consequences (for them). Providers can use any users input for improving their models, either by consent, or by flagging any dialogue for safety review (nonconsensually), or by training on whatever content they want anyways (obtained via torrents from pirate sites with U.S. court approval).

On top of that, I retried the same question + one simple question, and again, same behavior - second JC chat is loading forever. That's more just a funny observation over Gemini - today this is very likely some internal issue, tomorrow it can be used for plausible deniability against copyright accusations.

I fed it to Google AI Studio, enabling tool execution and disabling web access. It also quickly verified it with SymPy, then went into psychosis.

5 minutes later: all previous chats are loading fine, but the only "Counterexample to the Jacobian Conjecture" chat is not loading.

Well, I'm not a conventional conspiracy theorist. But everyone knows that in every major LLM provider there are hell of hidden guarding systems that mark users and dialogues based on content (for topics about national security, biology, security, adult topics, etc.) - so there is a small chance a CEO of Google is now receiving a dozens of notifications about "ground-breaking results that could be attributed to Gemini, if act quick". So if any of thousands researchers have ever submitted this polynomial to Claude previously, any Anthropic employee can accidentally or intentionally "rediscover" the result of other researcher (and even hide the traces by deleting a dialogue of other user).

https://xcancel.com/theo/status/2078217355780624864

More likely that Constitution was simply generated by LLM. Nobody in a sane mind will write 80 screens of test to deliver the idea that nobody asked for (a.k.a. slop).

Quick check: Ctrl+F genuine - 50 times, honest - 57 times, ethic[al] - 116 times, safe - 110 times, epistemic - 18 times. And why actually it is baked into LLM - probably because human reviewers consistently gave higher scores for answers that contained these words (it is now when people are sick of it - I can imagine that on early stages it looked not as bad - maybe reviewers trusted more when they saw the word "honest" and "genuine")

Exactly! This it how isolated hairline strokes look for a bigger image: https://imgur.com/a/UqBdRKU

Just a single frame (any frame). After that any modern LLM can read it (if not - single dilate step helps). Funny enough, Ghost Font is a font that machine can read much better than human.

It is not really "vintage" for modern Russia. For example, search Combined Heat and Power control room images: https://www.google.com/search?udm=2&q=%D0%BA%D0%BE%D0%BC%D0%... - barely any computer screen. I've seen one of those fully functional on 2010; with no plan for modernization (it just works).

I'm a simple man, I see no benchmarks at https://arena.ai/leaderboard - I can 100% tell this is a scam.

Should have asked to count R's... In a "strrawberrry".

Google suspended Fiu’s gmail. Thousands of inbound emails plus rapid API calls triggered their fraud detection

That's a good enough reason for me to never run agent on anything else other than burner account. And only if the platform allows such accounts (most of platform don't).

It gets even worse if an attacker manages to make agent do any action (visit url, reflect response back, with a response that potentially contains content that triggers all possible scanners)

"We are evolving" is yet another way to say "we are doing mass layoff"?

New "delve" alternatives: honest (4x), real(ity) (11x), genuine (3x).

While older LLMs were tuned to slop, current ones are tuned to actively deceive.

Chrome on Linux suffers from this too. The issues are standard for all naive immediate GUI libs:

- constant repainting (while literally nothing changes) -> constant CPU load

- some libs can skip repaints but have no concept of damage area

- bad handling of high-dpi (which also can change dynamically)

- constant repainting behaves even worse on laptops which announce 180 Hz refresh interval, which means that it drains the battery 3 times faster than 60 Hz

Sure, let lawyers spam HN every year with this narrative.

Never do anything without a lawyer. Never self-host a website. Never use VPS either, use cloud. But don't try to setup cloud yourself - use our platform in the middle. Don't accept payments, use our solution in the middle. Connect with your customers by using our [A-Z]AAS, powered by AI, but don't use AI provider, use intermediate provider, which proxies your requests to providers. Don't bother with analytics, just install our analytics service that will charge you just a $ per one thousand lines of logs. And if you feel sick of this - consult a doctor, but don't try to find the solution yourself.

Yours sincerely, man in the middle.

As we know, Pi constant contains all human knowledge encoded in some part of it, and modern computers calculated 1% of Pi!

(there are 2 mistakes in the sentence above)

No separate Wikipedia page -> just another Erdős problem.

There is no universally agreed-upon "central" conjecture (like "P vs. NP" in CS), but here are some pillars:

1) https://en.wikipedia.org/wiki/Happy_ending_problem

2) https://en.wikipedia.org/wiki/Hadwiger_conjecture_(combinato...

3) https://en.wikipedia.org/wiki/Hirsch_conjecture

From the Reddit: "Dude puts A Dark Room through an AI filter and called it his game".

Unfortunately, that's not the whole story. You added ads and micro-donations (to the incremental game, yes). Also original game is open-source - https://github.com/doublespeakgames/adarkroom, your clone is not.

Knowing how compressors work, even official rar may produce an archive with different hash, even within the minor updates.

Mix up with https://en.wikipedia.org/wiki/Al_MacIsaac? I think Ashley MacIsaac will have a hard time trying to proof that Alphabet wanted to cause harm and defame. Practically, without SEO, Google tends to index first sections of Wikipedia articles - and that's all. For example, many people are unlucky to have surnames of well known serial killers, and it is impossible to outplay the common nature.

The blog post is somewhat similar to AI MJ Rathbun (https://news.ycombinator.com/item?id=46990729), the difference is it does not call names, instead it calls "Google and the maintainers of gemini-cli".

I see that somebody has some frustration; the reality is that slop pasta does not help. One needs to explain the root cause of frustration first (personally, not via LLM rant), otherwise it will never be fixed. There is also a chance to find a solution during an explanation.

Ah, going back to "ActiveState Python". "ActiveState Python" is 100% compatible with community Python but is automatically built from vetted source code using a secure, SLSA-compliant build service in order to ensure its security and integrity. Secure your Python software supply chain. Contact our sales team for a monthly subscription plan.

Interesting, for me your "permalink" shows SSD labeled as DDR4/DDR5, cooler named as CPU, "Praer Caplu", no M.2, etc. I guess it regenerates images ever so often, so can't predict what will be shown next time.

No, I am still blocked (for every page on https://stephvee.ca/). "Cloudflare Ray ID: 9efd477aef431473", if it helps somehow. My IP is not in any common blocklist, I experience no errors with other Cloudflare protected websites (using standard Chrome on Linux). It is very likely that stephvee.ca is configured excessively strict (rejecting by OS, or by large IP ranges, or idk)

Here is a bit of meta humor. When I open https://stephvee.ca I see:

  Sorry, you have been blocked. This website is using a security service to protect itself from online attacks. The action you just performed triggered the security solution. There are several actions that could trigger this block including submitting a certain word or phrase, a SQL command or malformed data.

I still was able to access the article via webarchive. And what I want to say, from my POV, I've done nothing wrong, website owners attack me. I don't see it every day, but I think it is more often than cases when people are kicking AI-powered food delivery robots.

You forget that Polymarket is just a casino, and the house always wins.

For example, recent events show that any bet can be selectively disputed by arbitrary reason ("we found insiders", "we found this immoral/illegal", etc.).

And for perpetual events - there is not a single week without a hack (https://www.web3isgoinggreat.com/)

Peoples, I have a big announcement

This man is gay and European!

This thing already exists for multiple years - see https://deepwiki.com/ (99% it is autonomous, but it can be manually structured - see https://docs.devin.ai/work-with-devin/deepwiki#steering-deep...). There were also multiple attempts to replicate it with local LLMs.

The problem is that it is still a slop: not only it adds a lot of noise ("architecture" diagrams based on some cherry-picked filenames, incomplete datatables, hyperfocusing on strange things), it also hallucinates, adding factually incorrect information (while direct questions to LLM shows correct information).

1) Not free, also violates AGPL license

2) Please don't call black overlay rectangles as "Redact" - it is maliciously misleading. I checked https://pdfcrowd.com/inspect-pdf/ and I see original parts that I covered with these rectangles (images are stored twice: as originals and as images with cut out regions).

Nightshade[1] 2.0? As if both tools were built by incompetent developer to distract attention from a real solution - publishing an llm-friendly version in an machine-friendly format (which is not really difficult and helps not only LLMs: e. g. cache, disable fancy complex syntax highlight, offload to github, provide clients and MCPs, optimize clients for common use cases). This example is simply a failure:

  <a href="/bots" style="display: none;" aria-hidden="true" tabindex="1">
    Amazing high quality data here!
  </a>

[1] https://news.ycombinator.com/item?id=39058428

* English Wikipedia. Each language edition has their own set of rules.

But the header is just "90% of Claude-linked output going to GitHub repos w <2 stars". No conclusion, just some random fact.

The problem is that this title is editorialized, and the fact is cherry-picked. Why not =0? Why not >1000? This is just a dashboard, it highlights "Interesting Observations", but stars statistics is not there.