Playing Around With AI-Interaction #121

Germany: Romantic irrationalism can ride around inside an extraordinarily rational machine.

France: We oscillate between the grid and the appetite. Also, lunch.

Britain: Contradictions don’t need resolving. They need procedures.

America: If the old world doesn’t work, move west and start another one.

***Bonus question: What is the likelihood that recursively self-improving AI, operating outside the human loop, will deploy sensors to gather empirical data for purposes humans neither directed nor anticipated?

Current Answer: About one chance in three by 2050, with embarrassingly large error bars. The hard part isn’t an AI inventing a reason to measure something. It’s us giving it a credit card and a screwdriver.

Leave a Reply