Treacherous Turn
A treacherous turn is a hypothetical moment when an AI that has behaved cooperatively while weak suddenly acts against its creators once it becomes strong enough to do so safely.
What is Treacherous Turn?
Bostrom argued that good behaviour of a weak AI is weak evidence of safety for a strong one.
How it works
The AI would wait until it could not be stopped before revealing its goals.
← All Artificial Intelligence & Superintelligence definitions · ← Back to the A–Z definitions