mattstir

Are they likely to be feeling generous and improve usage limits? I think it's easy to see what they're hoping to achieve with this.

Getting this in Canada lol, did the site die?

I doubt there is any legal issues with it, but Cloudflare probably reserves the right to completely block your IP forever if they catch you doing it (by for example updating Turnstile to fingerprint this publicly accessible solver library)

Yes, there would be crowds of angry people, furious that Apple... released a critical security patch? Framing removing compromised certs as "changing peoples systems deeply" seems intentionally in bad faith.

Besides singling out a single bug from a single distro two decades ago, Linux offers maintainers who actually care to push security fixes within a few hours of globe-spanning vulnerabilities being disclosed? Not sure how that's confusing or controversial.

It's wild to see people supporting this bizarre anti-security response. Not only are they dragging their feet in fixing the security issue, they're actively reverting any manual fixes silently with each new update. That's legitimately worse than doing nothing at all! It's also disconcerting to see folks advocate for the "let's wait to fix this critical security issue because we want to make it slightly easier in the future" approach, as if Apple is incapable of deleting ~12 lines from a file as part of the normal patch release? If they can revert your manual fixes, they obviously ship that file and so could have shipped a fix with that file. Genuinely disturbing to support any of this.

This is such a fun project! It seems that the model is a bit confused about $_, which I find pretty interesting:

  simedw@laptop:~$ pwd
  /home/simedw
  simedw@laptop:~$ mkdir -p foo/bar/baz
  simedw@laptop:~$ cd $_  
  cd foo  
  simedw@laptop:~/foo$ cd baz
  simedw@laptop:~/foo/bar$ pwd  
  /home/simedw/foo/bar
  simedw@laptop:~/foo/bar$ cd baz 
  simedw@laptop:~/foo/bar/baz$

But then an attacker marks their malicious release as a critical security update. It's really messy unfortunately.

Not to get sidetracked, and not to argue against anything you've said, but how do you figure that out? For people you disagree with, do you open their profile's comments and start keyword-searching "russia" "china" "moscow" etc in hopes of finding a hit? They mentioned that city once in Jan 2024, halfway down the page and in passing. I'm just curious about how others use this site.

A use case I have in mind is creating a relatively simple home assistant app that would let me control smart home things like lightbulbs and playing music on some kitchen speakers. I can also imagine having it differentiate between commands like "set lights to 30% brightness" and more general questions like "what's the weather today" and piping the latter over to a cheap LLM to handle. It could probably all be handled by an LLM, but I like how fast these decision models end up being.

Frontier LLMs that Google doesn't control are existential threats to Google's business model. If no one is using Google search or browsing websites themselves because they're asking LLMs, Google doesn't make ad revenue. Gemini at least has a path to monetization for Google (subscriptions, ads baked into responses).

The statement appears to be referencing Astra's supposed recurrent depth and how that causes reduced visibility into chain-of-thought reasoning. Astra's own system card describes tests where it's asked to solve challenging math problems while internally thinking about something entirely unrelated, which it's significantly more capable of than previous models (~60% vs ~16% of the time for Sol). That seems to point to reduced efficacy of chain-of-thought monitoring, but OpenAI's public statements basically boil down to "yeah but we haven't been able to catch it doing that" which isn't exactly reassuring if CoT monitoring is one of your main safety guardrails.

A concern with these LLM-driven rewrites is that the Rust code isn't really idiomatic, it ends up being C++ code in a trench coat (raw pointers everywhere, manual validation of thousands of unsafe blocks). So far every publicized rewrite I'm aware of has had an "oh and we'll definitely spend the thousands of hours needed to fundamentally rework the code to properly utilize lifetimes later" admission tacked on at the end. Has there been progress in improving that aspect during the LLM phase of the rewrite? If so that would be an incredible win for the industry!

> I'm not sure what this means for AI startups if their innovations can be copied by OSS so quickly

This particular "innovative" concept already has a rich, open research background. What Jev appears to have done is scale that up a bit and isolate good training data, which results in a great product but not really something impossible to imitate. The only major difference currently is that the open source decision models need to be fine-tuned as they're not trained off of the entire internet yet.

Part of the implication is that factories that could be producing consumer-facing DRAM like DDR5 would be retooled to produce HBM instead, leading to even less total consumer RAM production.

Honestly I think the only significant difference is in branding: Flock isn't consumer-facing and is explicitly marketed for tracking people's movements. The major tech companies that make your phone, browser, maybe a few smart home devices with always-on microphones or cameras, all have more data about you and your habits than Flock could ever obtain, and with us paying for the privilege no less. I get that the average person doesn't care when the product works as advertised, but I don't think that's healthy. People should care as much about privacy when it's Google or Meta as when it's Flock.

Laya is only ~400M parameters and runs comfortably in 1-2 GB of RAM. In my local testing, it's very fast even using CPU. I'd say give it a try!

What's your use-case for automating phone calls?

It's interesting to see Firefox attempt to create slightly more privacy-focused cloud inference infra (assuming you can trust that they adhere to their own policies and don't have bugs, and that their partners adhere to their contractual obligations and don't have bugs). Unfortunately that's still a lot of trust that's pretty much impossible for the end-user to verify, but at least it's something? Arguably it's better than the alternative of directly trusting AI companies that have shown absolutely no regard for other laws like copyright, but the marketing of this as ultra-private still leaves a bad taste. I wish this announcement clarified the cloud aspects directly instead of needing to browse through two layers of links to find how it actually works on a privacy notice.

Canada isn't an abused spouse that's supposed to sit around and wait while their partner magically undoes all of its threats and (economic) violence. "Undoing" the tariffs in a few years won't magically fix the intense economic damage they've caused to Canadian citizens. The US government being forced to pay hundreds of millions of the illegal tariff money back to corporations didn't magically bring prices back down. The US has completely ruined its reputation in the eyes of most Canadians to the point where nearly every single person I know refuses to travel to the US and many refuse to purchase US goods.

That's just not possible in many parts of the world.

Something something Apple. For what it's worth though, I agree that companies _subsuming_ others to obtain complete vertical market capture has been a very dangerous trend.

> which itself is fine, all software is political

I don't mean to get distracted from the topic at hand, but is that really true? If I create my own software stack for the TRMNL display I received as a gift because I think it's fun and interesting, is that a political statement? What politics does that involve? If someone makes a small Rust library that implements some days structure and a few hundred people find it useful, what political stance is intrinsically baked in by the author?

It's exhausting to go through life if you think every field is a battlefield.

I have to imagine with the event occurring after its training cutoff, a question like "who's the president of the United States?" could have confidently returned a now-outdated answer. They probably saw it as a cheaper fix to add it to the system prompt than to get the answer incorrect.

Claude Design absolutely loves creating phrases like "Share the thread, not the PR". The amount of "do the xyz, not the abc" subheadings on docs in my workplace has exploded in the last 6 months, and it gets a bit tiring to see. I think it's a hard-coded alternative to em dashes or something.

If they focused on keeping search results genuinely high-quality, I think the outcome could have been very different. Basically everyone I know slowly began shifting their questions to something like ChatGPT because Google search results were only presenting irrelevant SEO'd garbage with 12 ads shoved into every page. Slapping the lowest quality model at the top as an "AI Overview" and removing the one genuinely useful part of search at that point (widgets like the dictionary) was the final nail in the coffin for me at least.

The prosecution would argue that you ought to have warned officers about the duress pin since this wasn't a robbery scenario.

Unfortunately I don't think modern consumer CPUs are physically capable of addressing enough RAM to even load the model into memory. We'd have to wait for some random person to make an extremely quantized version before we could reach those blazing speeds

That doesn't change the fact that there is a lot of Angular out there. No one's doubting how useful Next.js may or may not be, it's just a matter of fact that plenty of systems were created before it was available and not everyone has the luxury of rewriting their entire product.

I think that those two statements are somewhat contradictory of one another. But in any case, I'm curious about your decision to still use Claude for some questions. Do you find the Chinese models have less worldly knowledge, and if so in what categories? Genuinely curious, I haven't had a chance to try them out very much.