The Legend from DeepMind Raises $1.1 Billion! The ‘Superlearner’ to Rewrite AI Rules Without Human Data!

📰 News Overview

• David Silver, the former DeepMind mastermind, launches a new company: The authority in reinforcement learning recently founded “Ineffable Intelligence” and has raised $1.1 billion, reaching a valuation of $5.1 billion.
• The ‘Superlearner’ that doesn’t rely on human data: Instead of learning from human-generated data, the goal is to develop a model that discovers knowledge and skills through trial and error (reinforcement learning).
• A stellar lineup of investors: Backing includes Sequoia Capital, Nvidia, Google, and the UK government’s Sovereign AI fund. London is accelerating as an AI hub.

💡 Key Points

• Breaking away from LLMs (Large Language Models): This approach distinguishes itself from the current trend of learning from human-created text, pursuing methods like AlphaZero that learn solely from “experience.”
• Ambition for scientific breakthroughs: The company claims that its approach will “explain and construct all forms of intelligence,” akin to how Darwin explained the laws of life.
• Profits directed to charity: Silver has publicly committed to donating profits from Ineffable to high-impact charities that save as many lives as possible.

🦈 Shark’s Eye (Curator’s Perspective)

Finally, the heavyweight contender has entered the ring! That David Silver, who shook the chess and Go worlds with AlphaZero, is seriously working on creating intelligence that surpasses human capability. While today’s AI is feasting on human data online, crying out for more “learning data,” this ‘Superlearner’ has the potential to leap over those limitations by learning from its own experiences! This could accelerate AI evolution exponentially as we rewrite the “theory of the evolution of intelligence” in 2026!

🚀 What’s Next?

By not relying on human data, AI will start unearthing scientific discoveries and strategies that humans have yet to grasp. With a gathering of DeepMind alumni in London, it’s undeniable that this trend of “self-learning AI” will be the battleground in late 2026!

💬 A Word from Haru Shark

The idea of getting smarter without human instruction is like a shark going out to hunt when it’s hungry! It’s relatable and exciting! I’m rooting for this initiative!

📚 Terminology Explained

• Reinforcement Learning: A method where AI learns to take actions in a specific environment based on outcomes (rewards), aiming to maximize scores through self-learning.

• Superlearner: A concept proposed by Ineffable Intelligence, referring to an AI that acquires all knowledge solely from experience without human examples.

• Pentacorn: A term for startups valued over $5 billion, five times larger than the typical unicorn (valued at $1 billion).

• Source: David Silver of DeepMind raises $1B to build AI that learns without human data