Ethan Mollick
Generative artificial intelligence behaves less like traditional software and more like an unpredictable digital companion, requiring people to learn a new form of collaborative intelligence for work and learning.
The jagged frontier of capabilities means artificial intelligence performs unevenly, excelling at complex work while failing at simpler tasks in ways that can only be mapped through direct, personal experimentation.
Effective human-machine collaboration requires adopting either a centaur approach that strategically divides and delegates tasks or a cyborg approach that deeply integrates digital assistance directly into the human creative process.
Using the metaphor of treating the system as a person by assigning it a specific persona helps tailor the output and establish useful constraints, yielding less generic results than standard software prompts.