Partnering with Ineffable Intelligence: A Superlearner for the Era of Experience

2026-04-27 · Source: Sequoia Capital · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems · Depth: Advanced, quick

Summary

Sequoia Capital announced its partnership with David Silver and Ineffable Intelligence, a new London-based AI research lab, on April 27, 2026. Ineffable Intelligence's mission is to achieve superintelligence by developing a "superlearner," an AI system that acquires all knowledge solely through its own experience and actions, without pre-training on human data or imitation. This approach, rooted in Reinforcement Learning (RL) and guided by the "Era of Experience," aims to enable the system to rediscover and transcend human inventions like language, science, and mathematics. David Silver, known for leading DeepMind's Alpha series (AlphaGo, AlphaZero, AlphaStar, AlphaProof) and pioneering self-play in Go, is spearheading this effort. His work at DeepMind achieved superhuman performance, notably increasing AlphaGo Zero's ELO rating from ~3,700 to 5,000+ by removing human pre-training. Sequoia is co-leading Ineffable's initial funding round, supporting this ambitious and contrarian scientific mission.

Key takeaway

For AI Scientists and Directors of AI/ML evaluating long-term research strategies, this announcement signals a significant investment in pure reinforcement learning. You should consider exploring research paths that prioritize self-play and experience-driven learning over extensive human data pre-training. This contrarian approach, exemplified by David Silver's work, offers a transformative route to advanced AI beyond current LLM paradigms.

Key insights

Superintelligence may emerge from AI systems learning purely through self-experience, unconstrained by human data.

Principles

• Learning from first principles can transcend human knowledge.
• Self-play drives superhuman performance in complex domains.
• Pure experience-based learning avoids human data limitations.

Method

Ineffable Intelligence is building a Reinforcement Learning-based "superlearner" that discovers knowledge through its own actions and consequences in a designed environment, without pre-training or imitation.

In practice

• Develop agents learning from clean, environment-driven data.
• Explore self-play for complex problem-solving.
• Design environments for continuous, experience-based learning.

Topics

• Reinforcement Learning
• Superintelligence
• Self-play
• AlphaGo
• AI Research Lab
• Experience-driven AI

Best for: Research Scientist, AI Scientist, Director of AI/ML, Investor

Related on AIssential

• DeepMind’s David Silver just raised $1.1B to build an AI that learns without human data — Reinforcement learning can enable AI to discover knowledge and skills independently of human data.
• From games to biology and beyond: 10 years of AlphaGo’s impact — AlphaGo's victory in Go catalyzed modern AI, demonstrating systems can surpass human intuition and accelerate scientific discovery.
• Deepmind veteran David Silver raises $1B seed round to build superintelligence without LLMs — Reinforcement learning with world models offers a path to superintelligence beyond current LLM limitations.
• SKILL.MD is Not Enough — Integrating "experiences" alongside "skills" significantly enhances multimodal AI agent performance and strategic tool utilization.
• AI's Secret Footprint Exposed — Reinforcement learning, focused on goal-driven interaction with the world, is foundational to true intelligence, unlike large language models.
• [P] Vibecoded on a home PC: building a ~2700 Elo browser-playable neural chess engine with a Karpathy-inspired AI-assisted research loop — An AI-assisted research loop can rapidly develop high-performance neural chess engines on consumer hardware.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Sequoia Capital.