azinman2

20 ms·

391.
by azinman2 2y ago
I agreed with you until the last bit. I don’t think there’s a uniform “they” with a stable set of goals. All of the heritage foundation, Vance, Trump, Elon, congress, the Christian nationalists, etc I’m certain have different and probably o
392.
by azinman2 2y ago
Is there elsewhere?
393.
by azinman2 2y ago
Here’s my tip: Messages stuck with a badge but you have no idea what’s unread / how to clear it? Ask Siri for your unread messages. It’ll go through them and remove the badge.
394.
by azinman2 2y ago
What are you doing with it? For me deep research tasks are where 5 minutes is fine, or something really hard that would take me way more time myself.
395.
by azinman2 2y ago
It’d be great if someone would do that with the same data and prompt to other models. I did like the formatting and attributions but didn’t necessarily want attributions like that for every section. I’m also not sure if it’s fully matching
396.
by azinman2 2y ago
> For enterprises, AI directly replaces human labor with technology. It's cheaper, faster, more reliable — and often even outperforms humans. That’s… quite the claim. I guess we’re picking the worst people, the best voice-based AI,
397.
by azinman2 2y ago
Now. It’s amazing to me that everyone is like fuck OpenAI deepseek is the savior, when OpenAI’s papers and code jump started an AI revolution just a few years ago. Let’s wait the same number of years and see what deepseek does.
398.
by azinman2 2y ago
2025-01-30: Hirsch asked for an update as to whether clients running vulnerable systems have been alerted (no response as of publication) 2025-02-14: CVE-2025-26793 assigned 2025-02-15: publication So two weeks after they don’t respond what
399.
by azinman2 2y ago
I don’t hear about it anymore. What happened to it?
400.
by azinman2 2y ago
Shining a spotlight on an issue is completely different than the issue already existing.
401.
by azinman2 2y ago
No. DeepSeek and now related projects have shown it’s possible to add reasoning via SFT to existing models, but that’s not the same as a prompt. But if you look at R1 they do a blend of techniques to get reasoning. For Anthropic to have a h
402.
by azinman2 2y ago
To me the biggest surprise was seeking grok dominate in all of their published benchmarks. I haven’t seen any benchmarks of it yet (which I take with a giant heap of salt), but it’s still interesting nevertheless. I’m rooting for Anthropic.
403.
by azinman2 2y ago
I wouldn’t assume there’s no way to self host — it just costs a lot more than open weights. Anthropic claims they don’t train on their inputs. I haven’t seen any reason to disbelieve them.
404.
by azinman2 2y ago
It’s extremely unlikely that everyone is copying in a few weeks for models that themselves take many weeks if not longer to train. Great minds think alike, and everyone is influencing everyone. The history of innovation is filled with examp
405.
by azinman2 2y ago
I’m mostly amused and surprised to see a drag race gif on a security substack. Not surprised at any of the rest of it.
406.
by azinman2 2y ago
I strongly disagree. You’re literally putting people’s lives and possessions at risk who have no knowledge of this. There are many alternative methods, from getting the government involved to giving a a very long lead time to the vendor bef
407.
by azinman2 2y ago
Don’t have a WSJ account so can’t read the article, and the archive link posted doesn’t work for me. It sounds like this was a greenhouse based business? What I’m excited by (perhaps naively so) are the startups using robots to zap weeds an
408.
by azinman2 2y ago
But he does. That’s the reality. So how would you rather he spend it?
409.
by azinman2 2y ago
I would say this is highly irresponsible of the researcher to expose this publicly. These are people’s homes, along with their PII and locations. The residents didn’t choose this system, their building just uses it. They don’t even know tha
410.
by azinman2 2y ago
I think what’s really happened is we get output that’s close enough to normal communications to “feel” human. I could say it’s all a giant trick, which it kind of is, but we’ve also gotten to the point where the trick it also useful for man
411.
by azinman2 2y ago
That’s perfectly in line with the common role and understanding of books.
412.
by azinman2 2y ago
> The field became flooded with people who didn't care about anything but the money. This is what ruined SF. All the people who made fun of me growing up for being into computers were all of the sudden working at Twitter as a produc
413.
by azinman2 2y ago
Rodney brooks would disagree with you: https://rodneybrooks.com/predictions-scorecard-2025-january-... > Now self driving cars means that there is no one in the drivers seat, but there may well be, and in all cases so fa
414.
by azinman2 2y ago
Would love if they put latency in this too.
415.
by azinman2 2y ago
I wish they would have compared to the r1 distills of qwen2.5
416.
by azinman2 2y ago
Why should it be? I’d personally be pissed if my book, which came from my own hard work and is sold per person, all of the sudden get subsumed by a general AI. Even worse if it is commercialized and I get nothing for it.
417.
by azinman2 2y ago
The real question is why anything in the STRING is getting interpreted at all. In databases ‘null’ is not the same thing as NULL.
418.
by azinman2 2y ago
You’re assuming they won’t follow in OpenAI’s footsteps. OpenAI published a lot for a while and truly changed the world, far more than deepseek has. Only time will tell. But I think it’d be a mistake to think that this is necessarily benefi
419.
by azinman2 2y ago
Depends on the specifics of your life. As a gay man, I’m much happier in 2025.
420.
by azinman2 2y ago
What’s the process look like?

More ›