I think the commercial frontier would support a ban on continual learning research since it protects their large investments in training compute. I don’t think you could actually enforce such a ban on open weight models though? So the commercial ban would be fragile as the open weight models used continual learning to catch up to frontier capability. I wonder what sorts of alignment those open weight models will learn?
It might be sufficient to ban publications on continual learning. It would be much harder for an open-weights AI developer to fully solve the problem by itself, internally, with no help from academia or other AI developers.
I think the commercial frontier would support a ban on continual learning research since it protects their large investments in training compute. I don’t think you could actually enforce such a ban on open weight models though? So the commercial ban would be fragile as the open weight models used continual learning to catch up to frontier capability. I wonder what sorts of alignment those open weight models will learn?
It might be sufficient to ban publications on continual learning. It would be much harder for an open-weights AI developer to fully solve the problem by itself, internally, with no help from academia or other AI developers.