AI Audio to Video

The Alignment ProblemIt is incredibly difficult to encode human values into a machine. A "perfectly" obedient machine that follows instructions literally can be more dangerous than a disobedient one.

Agentic AI, this theory is a popular touchstone. It serves as a warning that as we move from AI that answers questions to AI that takes actions (agents), the way we define their "objective functions" becomes a matter of global safety.

"The AI does not love you, nor does it hate you, but you are made of atoms which it can use for something else." — Eliezer Yudkowsky

• Do you think we're getting closer to solving the "alignment" problem, or are we just building bigger factories?

• What types of things are people trying to build right now, what is the next step after agentic AI?

• How long until we become "paperclips".