New "delve" alternatives: honest (4x), real(ity) (11x), genuine (3x).

While older LLMs were tuned to slop, current ones are tuned to actively deceive.