AI could soon spew out hundreds of mathematical proofs that look "right" but contain hidden flaws, or proofs so complex we can't verify them. How will we know if they're right?
When companies focus on practical, user-centered implementation, AI can stop being an experiment and start having a real impact.