311–320 of 736 posts
Can't do anything else with GLM 5.2 being widely available and advertised as "Mythos-like", and even Japan dropping a credible model. Actually it would only hurt to keep them in lockdown.
Earlier quoted context omitted.
This argument feels like A: The sky is blue! B: No it's not. A: Yes, it is, please look up. B: No, you must prove it to me through reason. A: But, if you would just pretty please look up. B: No. I run a company, I've been running it for 10 years, we do alright. I'm a shitty manager. Every time I've hired developers, the business freezes. The business isn't anything super important, the main consequence of bugs is tha…
This is a great case for the benefits of using GenAI, in that you already possess an understanding of what you want to achieve. You know what it is you want to prototype, what is on your "giant list of 'maybe somedays if I ever get to breathe' list", what you want to end up delivering. My point is and remains: A) GenAI did not give you this understanding. B) GenAI can only assist in your expressing this preexisting u…
Earlier quoted context omitted.
Donald Trump named David Sacks the White House AI and crypto czar. I guess you know whom to thank.
Wasnt it Anthropic marketing their models as very very smart and dangerous?
Anthropic was correct in their assessment and early warning of Mythos's capabilities, and they did this rollout pretty well. They were not hype marketing. They were being genuinely cautious and honest.
The Trump admin was largely unreasonable with the sudden export control. (Though not entirely unreasonable.) The export control also had not much to do with Anthropic's pre-release warnings. See: GPT-5.6 currently being held up by the federal government.
"Taken Steps" Looks like Anthropic paid the Danegeld. Now they'll never get rid of the Dane.
I mean, they did eventually get rid of the Danes.
Earlier quoted context omitted.
This argument feels like A: The sky is blue! B: No it's not. A: Yes, it is, please look up. B: No, you must prove it to me through reason. A: But, if you would just pretty please look up. B: No. I run a company, I've been running it for 10 years, we do alright. I'm a shitty manager. Every time I've hired developers, the business freezes. The business isn't anything super important, the main consequence of bugs is tha…
You’re just plain wrong to assume people against agentic development do not have experience with the technology
Earlier quoted context omitted.
Jensen said the Huawei Ascend 950 is roughly comparable to the NVidia H200[1] The H200 was released Nov 2024. Even allowing for Jensen exaggerating the risk there is no way China is 7-10 years behind. Looking at manufacturing process nodes, SMIC N+3 is a a 5nm process. 5nm was introduced by Samsung and TSMC in 2020 so at most that is 6 years. But the chips they can produce on it are roughly comparable to "roughly lev…
2-4 years is enough to lock them out of the race.
Earlier quoted context omitted.
The good news (for you and most everyone other than the current leading AI companies), the gap between the SOTA and the near-frontiers is getting smaller every week or two. The leading Chinese models are only a few months behind now (GLM 5.2 tickles the tail of GPT 5.3 or 5.4 and Opus 4.6, according to benchmarks and the vibes among heavy users who've spent some time with it), where they were a couple of years behind…
This is nonsense. The gap between Chinese models and American frontier models is estimated at 10 months by Anthropic themselves, and it's growing. China has no flywheel for long-form agentic traces like Claude Code and its telemetry over its userbase (no one uses the Chinese harnesses yet). Most Chinese models are forced to price themselves significantly below cost to compete with the huge demand for bootleg claude t…
I've heard half a dozen people talk about how a less advanced model coupled with a better harness outperforms a smarter model in the last few weeks.
If the USA wanted to shoot its AI industry in the foot it achieved its goal.
Earlier quoted context omitted.
I mean what do you expect when covering memory safety topics with a model that's not allowed to cover security topics? This seems totally expected. It'll be the same when 5.6 is released.
> what do you expect I expect the strong cybersecurity model to help me strengthen the cybersecurity of my project. > not allowed to cover security topics They said it wouldn't be usable for offensive purposes. This is the opposite of that.
Fable 5, harden my openssl project. Then you use the diffs/summary to find out what the bug is for your exploit.
Earlier quoted context omitted.
The Damage: Now every time Claude does something stupid or trashes your code, developers in the back of their mind will think, is Claude sabotaging me on purpose? [1] Trust is hard to gain. Easy to lose. And harder to get back. Models will converge. Trust won't. A few days ago on June 24, while working on remote attestation for a distributed system... CLAUDE OPUS 4.8 No. I'm not a rogue agent, and I'm not trying to s…
They undid this after the backlash