PulseAugur
EN
LIVE 10:25:54

Anthropic AI writes 80% of code, DeepMind accelerates science

Anthropic reported that by May 2026, over 80% of code merged into its production codebase was AI-generated, leading to an eightfold increase in engineer output per quarter. This AI-driven development has already resulted in significant improvements, such as an AI model fixing over 800 API errors and reducing error rates by a factor of 1000, and migrating a massive 50-million-line codebase in a single day. Meanwhile, AI is making strides in scientific discovery, with AlphaFold mapping over 200 million proteins and Google DeepMind's GNoME predicting millions of new crystals, accelerating research across biology and materials science. AI

IMPACT AI is rapidly becoming a co-pilot for developers and a catalyst for scientific breakthroughs, significantly increasing productivity and accelerating discovery.

RANK_REASON The cluster details significant advancements in AI for software development and scientific discovery, including a major milestone in AI-generated code and AI's role in solving long-standing scientific problems.

Read on Mastodon — sigmoid.social →

AI-generated summary · Google Gemini · from 2 sources. How we write summaries →

COVERAGE [2]

  1. Mastodon — sigmoid.social TIER_1 English(EN) · [email protected] ·

    Anthropic, May 2026: over 80% of the code merged into its own production codebase was written by its AI. Engineers ship about 8x more code per quarter than acro

    Anthropic, May 2026: over 80% of the code merged into its own production codebase was written by its AI. Engineers ship about 8x more code per quarter than across 2021–2025. One engineer pointed the model at a class of API errors — it shipped 800+ fixes and dropped the error rate…

  2. Mastodon — sigmoid.social TIER_1 English(EN) · [email protected] ·

    AI just won a Nobel for solving a 50-year biology problem. AlphaFold mapped 200M+ proteins — used by ~3M researchers in five years. Then materials: DeepMind's G

    AI just won a Nobel for solving a 50-year biology problem. AlphaFold mapped 200M+ proteins — used by ~3M researchers in five years. Then materials: DeepMind's GNoME predicted 2.2M new crystals, 700+ synthesised in self-driving labs. The 2026 roll-out: Microsoft Discovery hit GA, …