What AI Practitioners Could Learn From A 1989 MIT Dissertation