At today’s pace of development, superhuman AI, smarter than every human combined, could arrive in 2027 based on the best predictors’ median estimates.
Superhuman AI could lead to human utopia, human extinction, and everything in between. Use this tool to predict how likely you think each future is.
Once it’s built, there’s no turning back.
Should we build superhuman AI?
Based on what you know today, answer six questions in about a minute to see what you think the future looks like.
Before you see your odds
Every car ride carries a tiny risk of not arriving: about 1 in 10 million. We would not drive if the risk was much higher. Think about your risk tolerance for building superhuman AI in the same way.
If anyone builds superhuman AI, how sure would you need to be that this ends well (in utopia instead of oligarchy, dependence, extinction, etc.)?
Reminder: it is on pace to be built in 2027.
Your risk tolerance: you’d need to be –% sure building superhuman AI ends well.
If they build it, your odds it ends well: