OpenAI CEO Sam Altman is once again making headlines with his latest post on X. He shared a post predicting economic collapse in a post-AGI world hit by mass unemployment. AGI, or artificial general intelligence, is a theoretical advanced form of AI that matches or exceeds human cognitive abilities across all domains, including learning, reasoning, and planning.
Altman's Contradictory Statements
"Post-AGI, no one is going to work and the economy is going to collapse," Sam Altman wrote. In another statement, he announced switching to a polyphasic sleep schedule—a practice of sleeping during multiple periods over 24 hours—because there is too much happening in the AI world to miss out. He wrote: "I am switching to polyphasic sleep because GPT-5.5 in codex is so good that I can't afford to be sleeping for such long stretches and miss out on working."
The irony is hard to miss: the man building the tech that is supposedly going to end work cannot stop working because of it. Altman's two statements are deliberately contradictory—one echoing the doomer fear that AGI will render human work obsolete, and another admitting he finds the tech so compelling that he sacrifices sleep to use it more.
AGI Timeline and Industry Skepticism
In 2025, Altman predicted AGI could arrive by 2030—a view often questioned by other tech leaders. Moltbot creator Peter Steinberger recently distanced himself from AGI, calling for specialized intelligence instead. Speaking on the Y Combinator podcast, Steinberger argued that the industry's focus on AGI ignores how humans and technology succeed through specialization, according to Business Insider.
Steinberger is not alone. Anthropic president Daniela Amodei called the AI concept outdated, while Google DeepMind CEO said AGI cannot be achieved without "world models."
OpenAI Launches GPT-5.5 in Codex
Altman's post comes as OpenAI launched GPT-5.5. The new model brings significant improvements in reasoning, coding, and scientific research. OpenAI stated that GPT-5.5 is faster, more efficient, and safer than its predecessor. It builds on GPT-5.4 but introduces key upgrades, handling complex multi-step tasks with better planning, tool use, and self-correction. The model is particularly strong in coding and debugging, achieving 82.7% accuracy on Terminal-Bench 2.0, outperforming rivals like Anthropic's Claude Opus 4.7 and Google's Gemini 3.1 Pro.
Despite its increased intelligence, GPT-5.5 maintains the same speed as GPT-5.4 while using fewer tokens per task, making it more cost-effective for enterprises and developers.
