Co-Intelligence: Living and Working with AI
Source: https://www.penguinrandomhouse.com/books/741805/co-intelligence-by-ethan-mollick/ ↗
AI not as a tool but as a cognitive companion that transforms how you work, decide and organise.
Mollick, from Wharton, argues that AI changes the nature of work not because it performs tasks for you, but because it alters your relationship with knowledge, creativity and decision-making.
It is the most accessible and up-to-date account of the lived experience of working with AI — less economic theory, more practical observation of how habits, roles and expectations shift.
Central argument
Mollick argues that AI is best understood not as an automation tool but as a 'co-intelligence' — an entity that augments human cognition by changing how we access knowledge, generate ideas, and make decisions, rather than simply executing tasks faster. His central claim is that the most significant effect of AI is relational and epistemic: it reshapes the habits, expectations, and self-conception of knowledge workers. Drawing on his experience at Wharton, he documents how integrating AI as a genuine collaborative partner — not a search engine or a script runner — produces qualitatively different outputs and alters what expertise itself means in practice.
Critique
Mollick's account, grounded in the lived experience of a highly educated, analytically sophisticated user, may systematically overestimate how transferable these benefits are across different professional contexts and skill levels. The 'co-intelligence' framing assumes a user who already possesses enough domain knowledge to critically evaluate, redirect, and challenge AI outputs — which is precisely the capacity that novices or generalists lack. This creates a tension the book does not fully resolve: AI as cognitive companion may amplify existing expert judgment rather than democratise capability, which has significant implications for team design and hiring that go largely unexamined.
Why it matters for product
For a CPO, Mollick's argument has direct implications for discovery and decision-making processes: if AI changes the nature of expertise rather than just task throughput, then the question is not how many research or strategy tasks AI can absorb, but how it reshapes what product judgment looks like and who on the team can exercise it. This should prompt a rethink of role definitions — particularly around research, prioritisation, and synthesis — not as efficiency gains but as structural changes to how product intelligence is generated and distributed across the organisation. It also raises a harder question about metrics: if AI augments creative and strategic work, current productivity proxies almost certainly fail to capture where the real leverage is being created or lost.