Sequoia Capital announces its partnership with David Silver and his new AI research lab, Ineffable Intelligence, based in London. The lab's mission is to build a superlearner — a system that discovers all knowledge purely through its own experience using reinforcement learning, without any pre-training on human data. The vision is a system that could rediscover and surpass humanity's greatest intellectual achievements, from the laws of physics to new branches of mathematics, by learning entirely from the consequences of its own actions.

Alfred Lin frames this as a deliberately contrarian bet against the dominant LLM paradigm, which relies on training over vast amounts of human-generated internet data. Silver's track record lends credibility to the approach: he was the lead researcher behind DeepMind's Alpha series (AlphaGo, AlphaGo Zero, AlphaZero, AlphaStar, AlphaProof), where removing human pre-training and relying on self-play produced decisively superhuman performance in Go. The key insight is that AlphaGo Zero, trained without any human data, dramatically outperformed its predecessor — suggesting that human knowledge can be a ceiling rather than a foundation. Sequoia is co-leading Ineffable's first funding round, positioning it as one of the most ambitious scientific bets in the current AI landscape.

This Skycrumbs blog post rounds up the most significant AI research developments from July 2026, spanning reasoning models, long-horizon task completion, hallucination reduction, video understanding, structural biology, mathematical problem solving, and model efficiency. A key finding from MIT and Stanford researchers shows that self-correction during reasoning matters more than model size, suggesting smaller models could rival larger ones with the right training approach. DeepMind introduced "prospective credit assignment" to improve multi-step task completion, while AI2 published research linking hallucinations to underrepresented facts in training data, bolstering the case for retrieval-augmented generation.

Meta AI demonstrated progress in understanding 30+ minute video sequences, and multiple teams adapted diffusion models to predict protein dynamics rather than just static structures—a meaningful advance for drug discovery. Google DeepMind's math reasoning system scored in the top 1% on International Mathematical Olympiad problems, and ICML 2026 papers on selective activation sparsity showed models performing comparably to ones three times their size. The post also previews upcoming publications, including OpenAI's GPT-5 architecture details, Anthropic's work on scalable oversight, and Meta's Llama 4 paper, framing the second half of 2026 as one of the most active periods in AI research history.