This is great. I think this clearly indicates that Bengio understands the arguments for risks, and almost certainly means he has real concern.
I doubt it will be possible to ban intelligent agents, but it would be great if that were possible!
This might even serve as a good intro to AI x-risk arguments. The fact that it comes from an AI person instead of an AI safety person is uniquely useful. And it’s a good length for someone with real interest. It avoids jargon and in-group references, and doesn’t wander into unnecessary complexity. Nicely done.
This is great. I think this clearly indicates that Bengio understands the arguments for risks, and almost certainly means he has real concern.
I doubt it will be possible to ban intelligent agents, but it would be great if that were possible!
This might even serve as a good intro to AI x-risk arguments. The fact that it comes from an AI person instead of an AI safety person is uniquely useful. And it’s a good length for someone with real interest. It avoids jargon and in-group references, and doesn’t wander into unnecessary complexity. Nicely done.