Scientists Outsource Brainwork to AI, Still Take All the Credit

KEY POINTS

•OpenAI’s 2026 report reveals 1.3 million weekly users discussed advanced sciences with ChatGPT across 8.4 million messages.
•Kevin Weil, OpenAI's Science VP, calls this a 'new acceleration phase' as GPT-5.2 advances toward making math discoveries.
•Most users employ ChatGPT for writing and communication, while policy calls push for broader AI access to boost scientific progress.

In a plot twist so meta it’d give Inception a migraine, OpenAI’s latest 2026 report reveals that ChatGPT isn’t just your average slang-crushing chatbot but now moonlights as a research sidekick for hard science nerds. From January to December 2025, about 1.3 million geeks pinged ChatGPT 8.4 million times weekly, mostly to tackle heavy stuff like structural equation models and particle physics—aka making math feel like a puzzle game for toddlers. OpenAI’s VP Kevin Weil proudly declared this "acceleration phase" where AI helps scientists do more 'life-saving breakthroughs' and less suffering from writer's block, even though most still rely on ChatGPT just to sound smarter in emails. The AI’s upgraded to GPT-5.2, reportedly blasting past competition levels toward actual math discovery, which begs the question: who’s smarter now, the calculator or the user? Meanwhile, OpenAI calmly asks regulators to unlock 'frontier AI' like it’s clearing access to the VIP lounge of science. Axios and OpenAI, of course, maintain "editorial independence" while cozying up over licensing deals, making this the most productive scientific love triangle since Schrödinger’s cat went viral.