Training Data
Training Data

David Silver and Lasse Espeholt of Ineffable Intelligence: A Superlearner for the Age of Experience

October 13, 2026

AI Summary

5 min read

In 2009, David Silver wrote a PhD on reinforcement learning for the game of Go, a domain so complex that superhuman play was considered years away. He was wrong about the timeline — AlphaGo beat the world champion in 2016 — but his deeper conviction has only hardened: the path to superintelligence runs through systems that discover knowledge for themselves, not through systems that ingest everything humans have already written. Now, as co-founders of Ineffable Intelligence, Silver and Lasse Espeholt are betting the company on that conviction, building what they call a "superlearner" that starts from nothing and learns forever.

What a superlearner is — and isn't

A superlearner, as Silver defines it, has two properties. First, depth: it can begin knowing nothing at all and progress to discoveries that go far beyond what humans know. Second, breadth: it can be dropped into any stream of experience — simulated, digital, or physical — and just learn. The analogy is humanity itself. "We somehow started out knowing nothing at all and somehow discovered everything that humans know today," Silver says. "Everything we learned, we discovered it for ourselves."

Continue reading the full summary in the app — free to try.

Read Full Summary →

Free • No credit card required

Listen to Audio Summary Open in App

What you'll learn

  • 1 (00:00) **Opening & The Core Conviction** - Host introduces David Silver and Lasse Espeholt of Ineffable Intelligence, and David explains his long-held belief that learning from experience is the path to superintelligence.
  • 2 (03:57) **Why Start a New Lab?** - David explains why he left DeepMind to found Ineffable Intelligence, arguing his non-LLM path deserved a dedicated lab rather than being a corner of a product-focused organization.
  • 3 (07:09) **Lasse's Journey to Ineffable** - Lasse Espeholt describes his background scaling RL at DeepMind and Google Brain, and what drew him to co-found the company.
  • 4 (09:32) **Defining the Superlearner and First Contact** - David defines a "superlearner" as a system with incredible depth (discovering everything from scratch) and breadth (applicable to any stream of experience).
  • 5 (12:37) **Superlearner vs. Current LLMs** - David distinguishes his vision from current LLMs that make scientific discoveries, emphasizing full autonomy and freedom from human priors.
  • 6 (13:59) **Building the Superlearner: Crucibles and Milestones** - The practical work involves building a state-of-the-art RL stack and "crucibles"—simulated environments capable of supporting emergent complexity.
  • 7 (17:18) **The Reward Function: Simplicity Over Specification** - David explains their philosophy of using simple, easy-to-describe reward functions in complex environments to allow for emergent complexity.

+ Full timestamped outline available in the app

Show Notes

David Silver spent two years building AlphaGo on human data. When he removed human data altogether building AlphaGo Zero, the AI surpassed AlphaGo in under two days. He left DeepMind to run that same bet in pursuit of superintelligence. David and Lasse Espeholt, co-founders of Ineffable Intelligence, explain their plan to build a “superlearner”—an agent that starts knowing nothing and discovers science, mathematics, and language for itself by direct experience instead of through a pre-training run. Silver argues that human-language reasoning is a ceiling, not a foundation, that alignment via chain-of-thought monitoring is already reaching its limits, and that the science of how superintelligences coexist should be worked out in simulation before anything touches the real world. Espeholt describes the GPU cluster and the lineage from IMPALA and SEED RL that makes the experiment possible. Silver’s forecast: strong evidence within a year, superintelligence in three.


Hosted by Sonya Huang, Sequoia Capital



0:00 – Introduction

1:09 – What held true from Go to today, and what the LLM era changed

3:35 – Why leave DeepMind to start a new lab

5:58 – Why "Ineffable" (and the secret second reason)

7:03 – Lasse's path: scaling RL at DeepMind and Google Brain

9:31 – Defining a super learner and first contact

12:35 – Super learners vs. LLMs that make discoveries

13:59 – Day to day: the RL stack, crucibles and human-journey milestones

16:13 – Don't breadcrumb the objective: simple rewards, emergent complexity

21:16 – What a crucible looks like

24:14 – Go: one agent is boring, two is infinite

25:41 – Will agents rediscover language?

27:22 – Removing human priors: AlphaGo to AlphaZero

32:05 – Keeping human bias out of the crucibles

33:35 – Confidence, timing and super forecasting

36:24 – Is compute the driver?

38:13 – The ceiling on simulation and Rich Sutton's objection

43:44 – Alignment and the science of superintelligence

47:45 – Monitoring agents that don't think in English

51:01 – Pacing the frontier

52:54 – The compute setup: Vera Rubin on GCP

55:18 – Training, inference, simulation and RL scaling laws

57:40 – IMPALA and SEED RL

1:00:58 – 20 people and the compute-per-person ratio

1:04:49 – A research-led culture

1:05:47 – Predictions

Training Data

More from this podcast

Training Data →