2 weeks ago
19
ARTICLE AD BOX
OpenAI CEO Sam Altman is erstwhile again making headlines with his latest station connected X (formerly Twitter). Altman shared a station earlier contiguous wherever helium predicted economical illness successful a post-AGI satellite deed by wide unemployment.
For those unaware, AGI oregon artificial wide quality is simply a theoretical, precocious signifier of AI that matches oregon exceeds quality cognitive abilities crossed each domains, including learning, reasoning, and planning. "Post-AGI, nary 1 is going to enactment and the system is going to collapse," Sam Altman wrote. In different statement, Altman announced that helium is switching to a “polyphasic sleep” – a signifier of sleeping during aggregate periods implicit the people of 24 hours.
Why? Because determination is excessively overmuch happening successful the AI satellite to “miss out” by sleeping.Sam Altman wrote: “i americium switching to polyphasic slumber due to the fact that GPT-5.5 successful codex is truthful bully that one can't spend to beryllium sleeping for specified agelong stretches and miss retired connected working”. It seems similar Altman has made 2 separate, deliberately contradictory statements — 1 echoing the fashionable doomer fearfulness that AGI volition render quality enactment obsolete, and different admitting that helium personally finds the tech truthful compelling and productive that he's consenting to sacrifice slumber conscionable to usage it more.
The irony is hard to miss: the antheral gathering the tech that's supposedly going to extremity enactment can't halt moving due to the fact that of it.Notably, successful 2025 Altman predicted that AGI could get by 2030 – a presumption often questioned by different leaders successful the tech industry. Moltbot creator Peter Steinberger precocious distanced himself from the AGI, calling for specialised quality alternatively of generalised intelligence. Speaking connected the Y Combinator podcast recently, Steinberger argued that the industry's absorption connected AGI ignores the cardinal mode humans and exertion really succeed: done specialisation, according to Business Insider.
Steinberger is not unsocial successful his skepticism. Anthropic president Daniela Amodei called the AI conception 'outdated' portion Google DeepMind CEO said AGI cannot beryllium achieved without “world models.”
OpenAI launches GPT 5.5 successful codex
Sam Altman’s station comes arsenic the AI institution precocious launched its GPT 5.5 AI model. In the authoritative announcement, OpenAI said that the caller exemplary comes with important improvements successful reasoning, coding and technological research.
The institution besides stated that GPT-5.5 is faster, much businesslike and safer arsenic compared to its predecessor. GPT‑5.5 builds connected the capabilities of GPT‑5.4 but introduces cardinal upgrades. The latest exemplary tin easy grip analyzable multi-step tasks with amended planning, instrumentality usage and self-correction.
The exemplary is peculiarly beardown successful coding and debugging. OpenAI besides stated that it managed to execute 82.7% accuracy connected Terminal‑Bench 2.0, outperforming rivals similar Anthropic’s Claude Opus 4.7 and Google’s Gemini 3.1 Pro.Despite its accrued intelligence, GPT‑5.5 maintains the aforesaid velocity arsenic GPT‑5.4 portion utilizing less tokens per task, making it much cost‑effective for enterprises and developers.
