This gives MoE (Mixture of Experts) a whole new meaning, albeit a slightly darker one.
How much data until we can intelligently route hard and soft human expert mixtures?
How much data until we can intelligently route hard and soft human expert mixtures?