sigmoid10

6,048 karmajoined 5 ปีที่แล้ว

comments

3 วันที่ผ่านมาdiscuss

That is more of a problem with the insane discovery system in the US than with anything else. If you had worked in Europe, GDPR would have protected your personal data from being sent over, as the Credit Suisse case has shown. They had to scrub all personal data before transferring files to a US counsel.

3 วันที่ผ่านมาdiscuss

B1 to B2 is a big step. Bigger than A1-2 to B1. B1 is basically just understanding people who speak slowly and clearly and being able to participate in everyday life situations. So a very gradual progression. B2 is already being able to do professional work in that language. This is what will take considerable time and active study. But it is also not a requirement for naturalisation in most countries. And many natural born citizens also never reach C1-2 level of proficiency either. So if you're B2, you may as well be counted as fully integrated.

3 วันที่ผ่านมาdiscuss

GDPR guarantees a right to privacy even on work devices. I think you need to filter out personal messages if compliance requires logging.

4 วันที่ผ่านมาdiscuss

I've lived in foreign places for less than 6 months and still bothered to pick up enough language to at least converse on a basic A1 level. Especially in certain regions where people don't speak english well this is almost a requirement for any daily life that is not work related. I can't imagine living somewhere for six whole years without picking up the language at all. Maybe if you actually hate the place and are sure you will leave again after that stint, but the above commenter doesn't seem to fall into this category either.

4 วันที่ผ่านมาdiscuss

But it does cover the state described in the top comment. A<-B is never going to be easily retrieved if you only experienced A->B during training, regardless if you are a human neural network or an artificial one. Also, you need to define "spatial" better. This is about logic after all and not geometry. Or topology? It's unclear which context you refer to. It's certainly not the topic of this thread.

4 วันที่ผ่านมาdiscuss

It certainly holds true for humans. The brain stores relational information in a sequential pattern that is not automatically reversible. One of the best examples is the alphabet. Everyone learns it in school, so the pattern A->B->C->... is trivial to recite for most people. Now, if I gave you a random letter to start with and asked you to to recite the remaining letters until Z, you'll probably find it is still pretty easy. But if I asked you to cite the letters backwards to A, most people would suddenly struggle with this task because they never learned or used the alphabet that way in school. You need to train specifically to link this kind of information backwards in your brain.

7 วันที่ผ่านมาdiscuss

If you look at public benchmarks like ExploitBench [1], then you'll see this is mostly a question of token budget. Once you give it sufficient tokens to burn, GPT 5.5 is roughly as good as Mythos when it comes to finding bugs and building exploits. With some clever auto-prompting to clear stalls, it even beats the base Mythos version. So Mythos' "magic" is not in the model, but in the harness and compute env. That's probably also why they never released it, because Anthropic already struggled heavily to make Opus available to the general public. Releasing Mythos publicly may well be technically impossible for them due to compute constraints.

[1] https://exploitbench.ai

8 วันที่ผ่านมาdiscuss

Youtube channels are also getting hyper-monetized now. Private equity firms finally learned that some of these informational channels draw a huge crowd of loyal viewers with a very specific kind of technical interest and have built high levels of trustworthiness. Ideal targets for running ads. Now they buy all these channels and have their marketers optimize every corner for generating easy money.

8 วันที่ผ่านมาdiscuss

Maybe project managers will finally get to experience flow states?

8 วันที่ผ่านมาdiscuss

This is not some fundamental problem. There are amazing options that don't suffer this downside. Most people are just too lazy and want something that works out of the box, so they accept or ignore the inherent risks of tools like Claude Code or Codex. But you just have to invest a little bit of research into what you need and which open model and harness suits your requirements. Then you can have your cake and eat it too. Long term this will be the standard anyways, it's only difficult while the frontier is moving so fast and aimed at ridiculous IPOs.

8 วันที่ผ่านมาdiscuss

Also should note that this is a working paper and not peer reviewed. Working papers are circulated to invite comments and discussion. Until some experts confirm their methodology I wouldn't take any of this for granted.

9 วันที่ผ่านมาdiscuss

That was only true before AI overview. Now all that info they scraped from public websites and fed into their models will still be available without anyone ever leaving Google's services.

9 วันที่ผ่านมาdiscuss

Well, because of things like COPPA I imagine most companies wouldn't want to risk anything here. So unless you have a website that is somehow guaranteed 100% blocking all traffic from the US or American citizens, you may as well implement it. But very few (if any) public websites will fall into that category.

9 วันที่ผ่านมาdiscuss

If you linked to the actual source of the study [1] instead of a random blog only talks about the result, you would see the big banner that the authors put there noting that the study is horribly outdated. Current models do make developers faster.

[1] https://metr.org/blog/2025-07-10-early-2025-ai-experienced-o...

9 วันที่ผ่านมาdiscuss

>It is suspicious to me that "age assurance" is trending EXACTLY as AI agents become capable of autonomously operating

It is not, because your premise is false. This whole thing has been going on for as long as kids have been online. The early 2000s tried (and obviously failed) by using credit cards. The UK tried and failed last decade to ban porn for minors this way. AI tools are probably not even on the radar for the kind of politicians that keep pushing this.

10 วันที่ผ่านมาdiscuss

Because of the introduction of AI overview, click-through rates are dropping like crazy. Up to 70% of Google searches now end without any clicks to third party websites. So most search users already stay completely within Google's ecosystem.

10 วันที่ผ่านมาdiscuss

This case was about PriceRunner, a price comparison platform that was suffering from Google prioritizing its own platform in search results. Klarna just happens to be the owner.

10 วันที่ผ่านมาdiscuss

Valuations are not permanent. Amazon dropped 90% during the dotcom bubble. And there is always another financial crisis coming.

10 วันที่ผ่านมาdiscuss

I'm thinking more of EULAs. Even if Anthropic somehow wedges this into their TOS, it might still be illegal. For example, in many US states this could potentially be classified as consumer fraud. You can't just sell one thing and then secretly and intransparently turn it into something else before shipping it. And in the EU it might violate GDPR too.

11 วันที่ผ่านมาdiscuss

Probably because the upgrades to the collider are so significant that it will be called the HL-LHC afterwards.