{"version":1,"type":"story","url":"https://digestai.news/story/google-engineer-says-self-improving-ai-remains-far-from-reality","json":"https://digestai.news/story/google-engineer-says-self-improving-ai-remains-far-from-reality.json","markdown":"https://digestai.news/story/google-engineer-says-self-improving-ai-remains-far-from-reality.md","slug":"google-engineer-says-self-improving-ai-remains-far-from-reality","headline":"Google engineer says self-improving AI remains far from reality","summary":"Google engineer Haydn Belfield argues that self-improving AI is not yet achievable, despite automation advances. At Google DeepMind, AI handles over 75% of production code but still requires human oversight for 88% of changes. Only 11.9% of code changes are fully delegated to AI, up from 7.5%, with success rates dropping sharply for complex tasks across 1,389 internal Argon agent runs.\n\nBelfield’s findings contrast with warnings from over 20 AI researchers, including Geoffrey Hinton and OpenAI’s Jakub Pachocki, who argue AI could automate all AI research within years. DeepMind’s General Capability Index (GCI) shows steady but unspectacular growth—0.1 points per month—with no evidence of AI-driven acceleration. The team attributes recent gains in Argon’s performance to increased compute and organization, not self-improvement.","keyPoints":["Google DeepMind’s AI writes 75% of production code but needs human review for 88% of changes","Only 11.9% of code changes are fully automated by AI, up from 7.5% previously","DeepMind’s GCI shows no measurable acceleration in AI research from self-improvement"],"whyItMatters":"Belfield’s findings challenge hype around self-improving AI, showing automation remains limited to routine tasks. The debate with top researchers highlights divergent views on AI’s trajectory and risks.","category":{"slug":"research","name":"Research","url":"https://digestai.news/category/research"},"entities":{"companies":["Google","Google DeepMind","OpenAI"],"models":["Argon","Gemini 3.7 Flash","Gemini 4 Argon"],"people":["Haydn Belfield","Geoffrey Hinton","Yoshua Bengio","Jakub Pachocki"]},"firstPublishedAt":"2026-10-14T19:18:44Z","updatedAt":"2026-10-14T19:18:44Z","sourceCount":1,"hasPrimarySource":false,"sources":[{"outlet":"The Decoder","title":"Google says self-improving AI isn't anywhere close","url":"https://the-decoder.com/google-says-self-improving-ai-isnt-anywhere-close","publishedAt":"2026-10-14T19:18:44Z","type":"press","primary":false,"lead":true}],"sourceNotes":null,"discussions":[],"thread":null,"cite":{"text":"Digest AI, \"Google engineer says self-improving AI remains far from reality\", 14 October 2026, https://digestai.news/story/google-engineer-says-self-improving-ai-remains-far-from-reality","publisher":"Digest AI","title":"Google engineer says self-improving AI remains far from reality","datePublished":"2026-10-14T19:18:44Z","url":"https://digestai.news/story/google-engineer-says-self-improving-ai-remains-far-from-reality"},"generatedBy":"Written by Digest AI's editorial model from the linked sources; the sources are the record.","license":"Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse"}