Anthropic Just Shocked The World... AI SINGULARITY Is Here!

A 14-minute explainer from a faceless channel, uploaded on 18 October 2026. The news hook was Jack Clark, Anthropic’s co-founder, giving the IMF’s Per Jacobsson lecture in Bangkok on 17 October and telling the audience the singularity has begun. The headline number that spread with it was a US economy of $44 trillion by 2030. The video separates the two claims, and that is the useful part.

Clark’s evidence, as he gives it, is about capability. Humanity’s Last Exam went from about 8% in January 2025 to 67.7% for Claude Opus 5.5. FrontierMath went from 2% to 94%. On TheoPhysEval, a new set of 50 unpublished theoretical-physics problems, OpenAI’s GPT-6-Astra ended at 98% (49 of 50) after the physicists found errors in their own reference answers. The narrator credits that score to frontier models in general; it is Astra’s. From this Clark pictures roughly 10 million human researchers directing a billion digital scientists.

The $44T comes from somewhere else: Anthropic’s Economic Scenarios for Transformative AI (Sept 2026; Anton Korinek, Chad Jones, Szymon Sacher, Tess Cotter and Peter McCrory), which has an interactive explorer. It is a task-based model, in which AI takes over shares of the tasks in each occupation. Against a no-AI path to about $33.5T in 2030 (2025 dollars), the modest scenario adds 1.6% ($34.1T), the substantial one 8.3% ($36.3T), and the extreme one 32.4% ($44.4T). The narrator corrects coverage that read the extreme case’s 15.4% growth rate as “adds 15.4%”. What surprised him is that AI-driven science barely moves the result: ideas get harder to find, the stock of ideas rises 0.6% by 2030, and the authors say the model can’t produce a singularity. The $44T is AI doing existing work, not Clark’s billion scientists.

The extreme case is also a distribution shock. Knowledge workers’ wages fall 11.5% against the baseline while everyone else’s rise. Overall unemployment reaches about 11.9%, and the labour share falls from 60/40 to 45/55. The authors reckon a transfer of about 9% of GDP would make knowledge workers whole, and the narrator points to the China shock as a reason to doubt it would happen. A Morning Consult survey of 10,980 Americans for the report found most people expect something like the substantial scenario. Eighteen economists, Daron Acemoglu and David Autor among them, reviewed a draft without endorsing it.