David Silver and his team are building a system that learns from its own experience. It is the next logical step in a decade of research that has progressively stripped human priors from AI.
Ten years ago, in front of two hundred million people, AlphaGo played Move 37. Two millennia of Go scholarship had dismissed the move. It won the match. It also proved something the field had argued about for decades: that a system could generate knowledge its architects did not have.
The researcher who led that work is now building Ineffable Intelligence, and we’re co-leading its seed round.
David Silver’s career focused on steadily removing human priors from AI systems. Atari from pixels. Go without game libraries. Chess and shogi from rules alone, in 72 hours. Then, MuZero mastered a range of games without being given the rules of any of them. Then, AlphaProof reached medal standard at the International Mathematical Olympiad. Each step took these systems further from imitation and closer to discovery. Ineffable is the next step in that line of work.
Distillation and discovery
The defining engine of frontier AI over the past five years has been training on the accumulated written, visual, and symbolic output of our species, then refining that capability against human judgment. The result is a generation of systems of remarkable generality — we believe that the frontier labs building at its edge will continue to produce enormous value.
That paradigm has made AI very good at distilling what humans already know. Generating knowledge we do not yet have, such as new theorems or new scientific frameworks, is a different kind of problem. It requires a system that can learn from something other than what humans have already produced.
That is the premise of experiential learning. Agents generate data through interaction with their environment and learn from those interactions. Learning and data generation are a single loop. Human-generated data is finite. Experience is not.
What Ineffable is building
Ineffable is building a system designed to generate knowledge from its own experience. One that learns not by extrapolating from human examples but by acting in engineering environments and learning from the signals those environments return. The founding bet is that this approach can eventually produce the kind of knowledge currently locked behind human limits: theorems we have not yet proved, scientific frameworks we do not yet have language for.
We have not seen this paradigm pitched anywhere else with the clarity, technical depth, and research pedigree that Ineffable brings. David’s arc from DQN through AlphaGo, AlphaZero, MuZero, and AlphaProof is a decade-long argument that experiential learning can be pushed progressively further from the scaffolding humans provide. Ineffable is the ultimate frontier for that argument.
We believe that the people David has assembled around him matter as much as the research arc behind him. The team includes some of the strongest minds working in reinforcement learning and frontier systems today, many of whom have worked with him for years before choosing to build this with him. Founding teams and researchers do not gather around speculative bets. They gravitate towards the ones they believe they can make work.
Ineffable is based in the UK with global ambitions, and is hiring across research and engineering. If this is the kind of problem you want to spend the next decade on, find out more and reach out here.
Building this, safely
We have spent the past several years investing across the AI stack, and the lesson that has come through most clearly is that this technology’s trajectory depends heavily on who is building it.
Powerful systems belong in the hands of people who are serious about what they can do and what they might do.
David Silver has spent his career not only pushing the frontier of what these systems can do but thinking carefully about what it means to put them into the world. That disposition is not incidental to why we wanted to back this company. Systems that learn beyond what their creators taught them demand builders who have thought hard about what that means.
Ineffable is opening a second frontier, one that runs parallel to the LLM wave but doesn’t depend on human knowledge to advance. We couldn’t be more excited to partner with David and the team.
The first frontier taught machines what we know. The second will teach them what we do not.
The content here does not constitute an offer to sell or a solicitation of an offer to buy any securities or investment advisory services.
The views expressed are those of the authors and do not necessarily represent the views or opinions of Lightspeed. Other market participants could take different views. Unless otherwise indicated, the inclusion of any third-party firm and/or company names, brands and/or logos are for representational purposes and does not imply any affiliation with these firms or companies and also does not imply their endorsement of the views expressed by the authors.
Certain information contained herein is based on information from various sources prepared by third parties. While such sources are believed by Lightspeed to be reliable, neither Lightspeed nor its affiliates assume any responsibility for the accuracy or completeness of such information, and such information has not been independently verified by Lightspeed. For more details please see https://www.lsvp.com/legal.
Authors
