Our Y4H Superlearner this week for always being kind, polite and hardworking as well as “never giving up”! 🌟💫⭐️🤩

Hc that Skids the theoretician superlearner can do anything and everything EXCEPT moonwalk

Silver's SuperLearner concept differs drastically from current LLMs, as it does not follow a scaling law, but is supposed, through reinforcement learning and self-play (as with AlphaGo and AlphaZero), to independently formulate, test, and validate or reject hypotheses in order to generate new knowledge. The methodological approach may well be the right one, but I see this second-order cybernetics as an obstacle to achieving this goal. I also doubt a purposeful, autonomous AI agent and rely on Hans Jonas' critique in the 1950s and 1960s of cybernetic devices, which were also described as teleological.

A suggestion: That the idea of an AGI (and ASI) capable of generating new knowledge is unrealistic, because AI, as of now, is trained with data already consumed by humans and is therefore limited in ways unnoticed by human intuition. David Silver has recognized this and proposes a radically different approach with his SuperLearner, but even here I remain skeptical due to second-order cybernetics.

David Silver represents a different but related direction. April 2026: David Silver raises $1.1B for Ineffable Intelligence with investors like Google and NVIDIA. → Goal: A superlearner that discovers knowledge without human data. → Method: Pure reinforcement learning. Self-play and autonomous improvement. Trial, error, hypothesis, validation. AlphaZero is the clearest proof of principle: a system that improved through self-play and reached superhuman performance. The broader ambition is to build agents that learn from consequences, not just from imitation.

Ineffable Intelligence has emerged from stealth with a European-record $1.1bn seed round at a $5.1bn valuation. Founded by David Silver, the lab is bypassing traditional LLM constraints to build a "superlearner" that discovers knowledge independently of human data.

DEEPMIND'S DAVID SILVER RAISES $1.1B FOR AI THAT LEARNS ENTIRELY ON ITS OWN 🧠 David Silver, the former DeepMind reinforcement learning lead, has founded Ineffable Intelligence, a new AI company focused on autonomous learning. 💰 The startup has raised $1.1 billion in seed funding, which means investors are backing a system built around reinforcement learning rather than human data. 🔍 The idea is a “superlearner” that improves through its own experience, so the real question is how far AI can go when it stops depending on curated training sets. #AI #ReinforcementLearning #DeepMind #Investment

$1.1B raised. Zero revenue. $5.1B valuation. This is what the next AI bubble actually looks like. 1/ David Silver launches Ineffable Intelligence — Founded: 4 months ago — Raised: $1.1B — Valuation: $5.1B Europe’s largest seed round 2/ No product No revenue No distribution Just one promise: build a “superlearner” AI that teaches itself everything 3/ Backers aren’t retail — $Nvidia — $SFGV Capital — Lightspeed Venture Partners — Sovereign state capital This is coordinated capital deployment 4/ What they’re really betting on: Reinforcement learning → beyond human data Same foundation that built AlphaGo Now scaled to: science, language, intelligence itself 5/ This changes the $AI game: Current models → trained on human data Next models → generate knowledge independently If true → data advantage disappears 6/ The real signal isn’t the tech It’s the structure: — Governments funding AI directly — VC state capital merging — Pre-revenue valuations in billions This is strategic, not speculative 7/ But here’s the risk nobody prices: — Timeline = unknown — Commercialization = unclear — Compute cost = massive You’re funding research, not a business 8/ This is how bubbles form: Capital chases narrative Narrative chases possibility Reality arrives last 9/ Final reality: This is not a startup It’s a moonshot on artificial general intelligence And the market is already assigning it billions Follow @blackintus. We see what others miss.

Meet David Silver, who just raised $1.1B at a $5.1B valuation in the largest seed round in European history. - Studied Mathematics at the University of Cambridge - Earned a PhD in Reinforcement Learning at the University of Alberta - Co-founded video game studio Elixir Studios as CTO - Joined @GoogleDeepMind to focus on reinforcement learning - Published foundational research that helped define modern deep reinforcement learning - Led the creation of AlphaGo, the first AI to defeat world Go champion Lee Sedol in 2016 - Built AlphaZero, a system that mastered games from scratch through self-play - Became one of the most cited and influential researchers in artificial intelligence - Served as Professor at University College London, shaping the next generation of AI researchers - Spent over a decade at DeepMind building systems that learn without human data - Left DeepMind in 2025 to found @IneffableLabs, pursuing a new path to superintelligence through reinforcement learning - On a mission to build a superlearner that can endlessly discover knowledge and skills without relying on human data - Just raised $1.1B at a $5.1B valuation led by @sequoia with participation from @nvidia and others - The largest seed round in European history “If successful, this will represent a scientific breakthrough of comparable magnitude to Darwin: where his law explained all Life, our law will explain and build all Intelligence.” (Ieffable Intelligence) orbitsearch.com/david-silver…

💸 The guy who taught a computer to beat the world's best Go player just got handed $1.1 billion. His pitch: build an AI that doesn't learn from humans at all. His name is David Silver. He just left a senior role at @GoogleDeepMind — the AI lab @Google bought in 2014, the one that made AlphaGo. AlphaGo was the program that beat the world champion at the board game Go. Silver started a new British AI company called Ineffable Intelligence. He raised $1.1 billion before the company has shipped anything. Investors include Google, @nvidia , and @sequoia Capital — some of the most important names in tech. To get why anyone wrote those checks, you need to know there are two ways AI can learn. The first way is the way you've heard of. Tools like @ChatGPTapp and @claudeai were taught by being fed enormous amounts of human writing — books, articles, websites. They got smart by copying patterns from humans. The second way is what Silver is famous for. AlphaGo didn't learn Go by being shown how humans play. It just played against itself, millions of times, and figured out winning moves on its own. No human teachers. No human textbooks. Silver's bet is that the second way is the future of AI. He calls what he's building a "superlearner" — an AI that discovers things on its own, without needing human examples. Here's what makes this more than one company's bet. Two other senior AI scientists just did the same thing. @ylecun, a famous AI pioneer, raised over a billion dollars for a similar idea. @_rockt raised hundreds of millions for his own version. Three of the most respected AI minds in the world all left Big Tech in the same season. They all bet that the current way of building AI — the one ChatGPT and Claude use — is hitting a wall. Their argument is simple. There's only so much human writing in existence. We're running out. The AI of the next decade will have to teach itself. If they're right, the AI you use today is the last generation that learned from people. One more thing worth mentioning: Silver pledged to give any money he personally makes from this company to charity. The thing to remember: Today's AI copies humans. The AI Silver wants to build teaches itself. What would an AI built without human data even know about us? #AI #PlainEnglishAI #FutureOfAI #TechExplained #DeepMind

this guy wants ai to earn its knowledge "a superlearner that discovers all knowledge from its own experience ... without relying on human data"

DeepMind에서 AlphaGo를 만든 David Silver가 Ineffable Intelligence를 창업. 시드 라운드에서 11억 달러 조달, 기업가치 51억 달러. 핵심 thesis, LLM은 막힌 길 "지구가 평평하다고 믿던 시대에 LLM을 풀어놓으면? 영원히 평평하다고 믿을 것" 목표는 시뮬레이션 속 superlearner AI 에이전트를 시뮬레이션 환경에 넣어 스스로 목표를 달성하고 협력하는 법을 학습시키는 방식. 바둑판처럼 닫힌 세계에서 현실의 복잡성으로 어떻게 확장할지가 핵심 과제. wired.com/story/david-silver…