Renowned mathematician Terence Tao recently sparked a fierce online debate by suggesting that AI is "non-renewably mining" open math problems. As automated systems rapidly digest unsolved proofs, technical communities are pointing out a hidden cost: when AI handles the grueling labor of complex problem-solving, humans lose the economic incentive—and the practical experience—needed to deeply understand the space.
For early-stage founders, this isn't an academic curiosity; it's a direct warning about defensibility. Using AI to shortcut core technical challenges gets a product to market fast, but bypassing the hard labor means bypassing the deep domain mastery that actually creates a moat. If your team didn't struggle through the underlying problem, you don't possess proprietary insight—you just possess an output that any competitor can prompt into existence tomorrow.
For small backers putting £500 into early-stage rounds, the takeaway is simple: look past cheap AI speed. The most resilient startups won't be the ones using models to mine quick answers from existing data, but the ones using AI efficiency to push further into hard, defensible human innovation.