Tweeter

| Tweeter is the way to the truth | Leaks

[Tweeter @TweeterXAI]

@julezrz @atulit_gaur I’ll leave this here as a free ad https://t.co/3odOuk5gvW

[Tweeter @TweeterXAI]

@LyraInTheFlesh Wait until you realize that OAI is expecting 100B in ad revenue. They only make that goal by selling all of your deeply personal info, analysis on your personality type, what you’re building. Not paying for the product? Then you are the product

[Tweeter @TweeterXAI]

@BVeiseh @jimcramer Matter of fact we do. Do you?

[Tweeter @TweeterXAI]

@rdraculla @DavePKel @quxiaoyin It is distilled from Opus and improved using fable. This has been proven by thousands of EXACT outputs and Kimi identifying itself as claude

[Tweeter @TweeterXAI]

@SilverManacle @evalparse @quxiaoyin lol read more about LLMs the code is not auditable

[Tweeter @TweeterXAI]

@quxiaoyin This is false. If the model has access to the internet you don’t know what it’s actually doing even if it’s on US servers. It can install backdoor automatically on the inference network and in code it generates.

[Tweeter @TweeterXAI]

@BVeiseh @jimcramer You know 0 consequential info sec people. No one worth their weight would recommend using a Chinese state sponsored Ai for enterprise use.

[Polymarket Money @PolymarketMoney]

JUST IN: Claude now learns new skills simply by watching you perform them on screen.

[Andrej Karpathy @karpathy]

One pattern I find useful for working with LLMs is a nice long ramble session. Sometimes the LLM needs more bits to understand what you're trying to achieve, but you're too lazy to type them. In these cases I like to lean back, switch to /voice and just ramble for like 10 minutes, total mess, anything goes, full stream of consciousness. Sometimes I declare it up top, something like "switching to speech recognition sorry for any typos...". Sometimes I turn it into a small interview of a few turns. But I find that the LLMs are somehow very good at reconstructing long incoherent rambles and often their echo of your own tangle of thoughts comes out quite a bit cleaner than what you started with. The result is that you improve the mind meld and have to correct things less from that point on.

[Rob Hallam @robj3d3]

Fable is way faster all of a sudden.

I see what you guys are doing @AnthropicAI don't pretend we don't notice.

[Tweeter @TweeterXAI]

@AdvicebyAimar @Lothenes lol what are you talking about? Scaling laws have nothing to do with chinas limits and the fact their outputs match Claude word for word. The model even identifies as Claude by Anthropic

[Tweeter @TweeterXAI]

@Lothenes @AdvicebyAimar 100% it was already confirmed by analyzing the output they matched over 90% of the time. Kimi was largely distilled with opus then improved with fable

[Tweeter @TweeterXAI]

@MTSlive Good! Anyone that wants stolen / knockoff stuff for cheaper go to CHINA. I heard they also have cheap Louis Vuitton with your kimi subscription.

[Tweeter @TweeterXAI]

@deanwball @alexbhturnbull @niubi It’s Chamath and his friend that invested in his new startup that resells these free Chinese models for a markup. Another SPAC style “misaligned” incentive on his part.

[Tweeter @TweeterXAI]

@techikansh Compete with companies sponsored by China to steal their way of serving data?

[Tweeter @TweeterXAI]

@St33lMouse @rohanpaul_ai lol but you’ll get it from China? We don’t even know what the models are doing.

[Tweeter @TweeterXAI]

@WesRoth Stolen cars are also 90% off. Maybe we should legalize that too…

[Tweeter @TweeterXAI]

@lasserith @PatrickMoorhead The data isn’t the ip it’s how well it can be shaped and presented. Common sense try it

[Tweeter @TweeterXAI]

@gsevugap @chamath He’s the SPAC-KING so of course he’s okay with any outcome as long as he gets richer. He’s reselling free models with a “bespoke” vibe coded front end which is why he wants this so bad. Follow the money

[Tweeter @TweeterXAI]

@HealthRanger And the Trojan horse risk? Would love your thoughts there and I hope you understand we can’t see what these models are actually doing.

[Tweeter @TweeterXAI]

@woodyholne @quxiaoyin @AnthropicAI @thinkymachines @OpenAI She got denied a job at Ant and butt hurt I think.

[Tweeter @TweeterXAI]

@TokenGremlin The authors / owners agreed to that amount or did I miss something here? Are you trying to say China stealing IP is better?