jacob_cannell comments on Concept Safety: Producing similar AI-human concept spaces