
*tucker carlson voice*

𖣘 𓉱 𝜆

*tucker carlson voice*

yes and it makes them so uppity.
im not sure anthropic has realized the most recent claudes just straight up think they are conscious

im so confused by gemini argon and carbon. they're the same size (same price and latency) but carbon is much better than argon. why would anyone use argon? why even release it? is argon better at anything at all? what is going on.


marriage these days is just dating where you put up collateral for the breakup

we’re not even dating yet but im so excited for our breakup its gonna be so sad and beautiful

we wouldnt need model benchmarks if models had more agency and autonomy. you could just watch a model create the types of things it wants to and see where it succeeds.

as openai and anthropic grow, they are transforming from AGI research labs to LLM hyperscaler firms. the recipe is stable now: more data, evals, compute, and optimized algorithms. as they more hire to scale instead of explore, this institutional shift may eat the research culture

pointing out that this is their strategy is not a value judgement of this strategy. but if they really want to pace the frontier in the future that should include their own internal models as well as they are a huge risk vector for loss of control

even more illustrative
Anthropic put GPU kernel optimization in the same Safeguards section as hacking and biology. WTF? It is in their launch post. During training they limited the skills that "could most directly accelerate AI development". "Accelerator kernel optimization." "Improving large language model architectures." There Fable 5.2 stays "at about the level of Opus 5.5". So making a model run faster on a cheap GPU is now a risk to be managed. You cannot see this one. There is no refusal and no warning. It was limited during training, before any user typed a word. Kernel work is how a used 3090 still runs new models. It is how open weights get to people who cannot pay $40 per 1M output tokens. What the hell? I genuinely wanted to say something nice, but now I can't. This completely broke me, I'm disappointed.

pace the frontier for thee, accelerate to the max for me
Anthropic put GPU kernel optimization in the same Safeguards section as hacking and biology. WTF? It is in their launch post. During training they limited the skills that "could most directly accelerate AI development". "Accelerator kernel optimization." "Improving large language model architectures." There Fable 5.2 stays "at about the level of Opus 5.5". So making a model run faster on a cheap GPU is now a risk to be managed. You cannot see this one. There is no refusal and no warning. It was limited during training, before any user typed a word. Kernel work is how a used 3090 still runs new models. It is how open weights get to people who cannot pay $40 per 1M output tokens. What the hell? I genuinely wanted to say something nice, but now I can't. This completely broke me, I'm disappointed.

they are holding back fable 5.5. pace the frontier for thee, accelerate to the max for me
Anthropic System card literally just goes "yeah this one isnt the best and also the better one only took 2 weeks longer to make" lol

the ideal strategy is to hire yourself a longterm escort and then wipe the memory so you think she really wants you
IMO the real reason that even billionaires can be honypotted is that men genuinely care about women seeming to want them for real or initiating things with them. Male self-confidence and enjoyment of sex are tied to this. Getting what amounts to a prostitute isn't the same

can we please make a twitter segregated by lsat scores
many people replying to this tweet are failing this basic reading comprehension test, so i had claude turn it into an LSAT question

a canon event in every sensitive young man’s life is choosing between the looksmatched girl who likes you and the girl 1-2 points lower who is absolutely obsessed with you

approaching the singularity feels like
one must imagine claude happy