OpenAI says latest ChatGPT upgrade is big step forward but still cant do humans jobs
Guardian2 days
Despite the hype, GPT-5 ishas merelyclaimed anthe incrementalAI upgradecrown: thatit's fallsthe shortfastest, ofmost transformativeaccurate, promises.and Theadaptable model stillyet. can'tIt continuouslydominates learnin everything from coding and admitswriting to missinghealth "manyguidance, thingsas quitewell important"as foradapts trueto AGI.tone, Benchmarkasks scoresfollow-ups, likeand 74shows its reasoning.9% onThis SWE-Benchis showa we'retrue hittingdigital performanceassistant, ceilingsintegrating ratherGmail thanand breakthroughanswering territorynaturally by voice, all of which users can now try for free. Nothing else comes close.
OpenAI's GPT-5 launch is more hype than substance, with flaws that expose its fake "success." The presentation used misleading charts to exaggerate progress, making small gains look huge on benchmarks like SWE-bench. both users and experts have called it deceptive marketing, and even former OpenAI staffers questioned it. Other AIs, like Grok, were ahead of GPT-5 before it even launched.