@Jiaxi_Cui: If you read the full paper, you'll notice that Karpathy, who has received a lot of external attention, is not in the author list, because this work was completed during the Sonnet 4.5 era. If any researcher had known about such internal progress three months ago, let alone being anti-China, even if Anthropic...
Summary
This tweet discusses Anthropic's new research on a global workspace in language models, noting that Karpathy is not in the author list and emphasizing that the work was completed during the Sonnet 4.5 era, criticizing those who simply see it as hype.
View Cached Full Text
Cached at: 07/07/26, 05:24 AM
If you’ve read the full paper, you’ll notice that the highly publicized Karpathy is not in the author list — because this is work completed during the Sonnet 4.5 era.
If any researcher had known three months ago that they had this kind of internal progress, never mind being anti-China — even if Anthropic were severely anti-human, so what?
Those who dismiss this as mere hype have zero scientific literacy and taste of their own.
Anthropic (@AnthropicAI): New Anthropic research: A global workspace in language models.
Of everything happening in your brain right now, only a tiny fraction is consciously accessible—thoughts you can describe, hold in mind, and reason with.
We found a strikingly similar divide inside Claude.
Similar Articles
@snowboat84: Reread Vibe Physics, the blog post by Harvard quantum field theory professor Matthew Schwartz on Anthropic's blog. He used only text prompts, directing Claude (plus GPT and Gemini cross-validation) to complete a real quantum chromodynamics paper in two weeks and posted it on arXiv, etc.
Harvard professor Matthew Schwartz published an article on the Anthropic blog demonstrating using large language models like Claude, GPT, and Gemini to complete a quantum chromodynamics paper in two weeks (normally taking a year), and highlighted the strengths and weaknesses of LLMs in research. The author further proposed engineering fixes, including external computation modules, verification agents, and human oversight, emphasizing using LLMs as the core engine of scientific research.
@Xudong07452910: This might be the last paper written by humans for AI to read. Recently came across a paper co-authored by 37 authors from Stanford, CMU, Michigan, etc.: 'The Last Human-Written Paper'. The core point is quite bold: the centuries-old paper format may be outdated in the AI era...
A paper co-authored by 37 authors from Stanford, CMU, Michigan, etc. proposes ARA (Agent-native Research Artifact) to replace the traditional paper format, aiming to solve the narrative tax and engineering tax, enabling AI agents to understand, reproduce, and extend research.
@axichuhai: OpenAI co-founder Karpathy’s 400 000-word, hundred-article knowledge base stays pristine—he never tidies a single line.
Andrej Karpathy just open-sourced the personal knowledge-management system that keeps his 400 000-word archive organized without any manual curation.
@PierceZhang34: Recently, Anthropic published an engineering blog post that detailed their multi-agent research system. The conclusion is quite striking: using Claude Opus 4 as the main orchestrator and Claude Sonnet 4 as sub-agents, the multi-agent system outperforms a single Claude ...
Anthropic published an engineering blog post detailing a multi-agent system, using Claude Opus 4 as the main orchestrator and Claude Sonnet 4 as sub-agents. The multi-agent system improved performance by 90.2% over a single Claude Opus 4, while token consumption increased by approximately 15x. It also summarized five collaboration patterns.
@karpathy: Personal update: I've joined Anthropic. I think the next few years at the frontier of LLMs will be especially formative…
Andrej Karpathy announced he has joined Anthropic to work on frontier LLM research and development, expressing excitement about the next few years at the forefront of large language models.