Safety shouldn't slow you down

Kaplan rejects the idea that responsible AI means slower AI. His position is that safety belongs inside the velocity of progress, not in opposition to it.

That means merging technical innovation with ethical responsibility from the start, so powerful systems stay aligned with human values.

The case for a democratic approach

Kaplan argues that a democratic architecture for superintelligence is safer, faster, and more profitable than current approaches. He estimates it could cut the probability of human extinction by AI from roughly 20% to under one-tenth of one percent.

Those are his estimates, not settled consensus. But the design logic is clear: spread the intelligence and the oversight across many participants instead of concentrating both in one system.

Your clicks are training data

Kaplan wants society to engage with AI actively, not passively. Every online interaction helps shape what emerging systems can do and what values they absorb.

His practical advice is to stay mindful of how you behave online, keep your critical thinking switched on when you use AI, and follow the work you're intrinsically drawn to.