DeepMind’s David Silver just raised $1.1B to build an AI that learns without human data

You know how kids learn some things just by doing them, no teacher required? A toddler doesn’t read a manual on how to walk — they just fall down a thousand times until their brain figures it out. That’s basically the idea behind what David Silver, one of the people who built AlphaGo (the AI that beat the world’s best chess and Go players), is now working on. His new company, Ineffable Intelligence, just raised a staggering amount of money to build AI that learns by exploring and experimenting rather than by studying mountains of human-generated text and examples. Most AI today is like a student who reads every book ever written. This new approach is more like a scientist running experiments in a lab.

Why does this matter? Because the current way of training AI is hitting some real walls. You can only learn so much from copying humans — at some point, if you want to go beyond what humans already know, you need a system that can discover genuinely new things on its own. Silver’s track record here is worth noting: AlphaGo didn’t just learn from human games, it played millions of games against itself and discovered moves that surprised professional players. The new lab wants to take that same principle much further, potentially into science, medicine, and problem-solving that goes well beyond anything in existing textbooks.

So what does this mean for your wallet right now? Here are three grounded ideas. First, if you run a small business, keep an eye on AI tools for data analysis — as this kind of self-learning AI matures, software that spots patterns in your sales or customer behavior without needing you to set it all up manually will get genuinely useful and affordable. Second, if you’re thinking about where to spend time learning new skills, fields like drug discovery, materials science, and logistics are where this technology will land first — getting even basic fluency in those industries through online courses now puts you ahead when hiring picks up. Third, freelancers in research or consulting can start positioning themselves as the human translator between these powerful AI outputs and clients who don’t understand them — that interpretation skill will be in real demand because the AI might find the answer, but someone still needs to explain it to the room.

The companies spending billions today on how AI learns are essentially writing the job descriptions and business opportunities of five years from now — knowing what they’re betting on is genuinely useful intelligence.