AI Control Problem
The AI control problem is the question of how humans can keep control over AI systems that become more capable than they are.
What is AI Control Problem?
Stuart Russell's book Human Compatible (2019) proposes machines that are uncertain about human preferences and so defer to people. The control problem covers both alignment and containment methods.
How it works
Strategies include aligning goals, limiting capabilities, monitoring and maintaining the ability to intervene.
← All Artificial Intelligence & Superintelligence definitions · ← Back to the A–Z definitions