Mollick opens by arguing that the most effective way to work with modern AI systems like large language models is to treat them genuinely as collaborative partners — engaging in back-and-forth dialogue and iteration — rather than as simple, one-shot query tools expected to produce perfect output immediately.

Core argument he establishes:

• Mollick draws on his own extensive, hands-on experimentation with AI tools across teaching, research, and writing to argue that iterative collaboration produces significantly better results than single, one-off prompts
• He introduces the framing of "co-intelligence" — human and AI capabilities genuinely combined and complementary, rather than AI simply replacing human judgment and effort
• This opening framing sets up the book's practical guidance for how to structure this kind of ongoing collaborative relationship with AI tools effectively

This collaborative framing distinguishes Mollick's approach from both AI-skeptic and AI-maximalist perspectives, positioning genuine human-AI partnership as the productive middle path.