I think that sounds very reasonable, but unfortunately these models don’t know what they know and don’t. A small model that knew the exact limits of its knowled

I think that sounds very reasonable, but unfortunately these models don’t know what they know and don’t. A small model that knew the exact limits of its knowledge would be very powerful.