While the West talks, China builds – Qwen’s new AI model just launched, and it beats DeepSeek V3 on various metrics

Gemini

Sensationalist narrative VS reality

Are we delusional or is everyone else?

[AI Explained] Nothing Much Happens in AI, Then Everything Does All At Once. Strong rumours of an o3+ model from Anthropic

What actually matters

38.1% isn't AGI I guess

OpenAI developing AI coding agent that aims to replicate a level 6 engineer, which its believe is a key step to AGI / ASI

OpenAI will launch o3-mini "very soon" followed by full o3 in "February, March, if everything goes right", with AI agents in Q1 2025 enabling ChatGPT to perform computer tasks like form-filling and web browsing

Google's Gemini 2.0 Flash Thinking Exp 01-21 model now has a context window of over 1M tokens.

DeepSeek R1 on par with O1 Preview in ARC-AGI

Dario Amodei said, "I have never been more confident than ever before that we’re close to powerful AI systems. What I’ve seen inside Anthropic and out of that over the last few months led me to believe that we’re on track for human-level systems that surpass humans in every task within 2–3 years."

Can't outsmart DeepSeek V3 DeepThink

Introducing Kimi k1.5 --- an o1-level multi-modal model

R1 performs 11 points better than its non Chain-of-Thought counterpart on Livebench

Open source o3 will probably come WAY sooner than you think.

@btibor91 on X: OpenAI website already has references to Operator/OpenAI CUA (Computer Use Agent) - “Operator System Card Table”, “Operator Research Eval Table” and “Operator Refusal Rate Table” (preview of tables rendered using Claude Artifacts)

o3 and o3Pro are coming - much smarter than o1Pro

First Screenshots of Grok 3

OpenAI Senior AI Researcher Jason Wei talking about what seems to be recursive self-improvement contained within a safe sandbox environment

Noone I know is taking AI seriously

AI labs should resist to the pressure of handing ASI to fascists like Trump and Musk, just like Strassman did in 1939 in nazi Germany

AI predictions from Simon Willison: still no good AI agents 1 year from now, but in 6 years ASI causes mass civil unrest

The goalposts on what is AGI keep moving. Hope Deep Mind does a follow up of this paper and where we are today. I believe we are in the expert level now, what about you all.

Ethan Mollick:"There has been a shift in recent weeks where insiders in the various AI labs are suggesting that very intelligent AIs are coming very soon. [...] researchers inside AI labs appear genuinely convinced they're witnessing the emergence of something unprecedented"