Co-Intelligence: Living and Working with AI
by Ethan Mollick
Portfolio, 2024

Over the weekend, I caught up with a friend I hadn’t seen in years. When I explained that most of my work these days is focused on AI, he asked if there was one book he could read to understand how to better use AI. This is the book I recommended.

Why? Because it offers

  1. a clear explanation of how the technology works,
  2. thoughtful explorations of what it means for us, and
  3. practical suggestions for using it to help with common tasks.

AI isn’t like other technologies. As Mollick puts it,

We have invented technologies, from axes to helicopters, that boost our physical capabilities; and others, like spreadsheets, that automate complex tasks; but we have never built a generally applicable technology that can boost our intelligence.

The book is divided into two parts. The first explains the fundamentals, including how transformers work and the challenges inherent in aligning AI goals with human goals.

Mollick lays out four principles for working effectively with AI:

  1. Always invite it to the table: AI is a general-purpose technology; it behooves you to try it for different ends.
  2. Be the human in the loop: assume AI will need supervision and guidance; provide it.
  3. Treat AI like a person, but tell it what kind of person it is: don’t fall into the trap of anthropomorphizing AI – but it’s helpful to suspend your disbelief.
  4. Assume this is the worst AI you’ll ever use: the technology is improving fast; assume the things that are difficult or impossible today will be doable in the future.

The second part explains how to use AI effectively by seeing it through the lens of five possible roles:

  • As person: even though we should remember AIs aren’t people, it can be useful to ask it to simulate human attributes such as personality and particular expertise.
  • As creative: AI can assist us in our creative efforts or replace our creative functions outright. You’ll be better served by using AI for creative augmentation rather than replacement.
  • As coworker: AIs can handle information tasks at much greater speed and scale than people. Knowing what they can and can’t do well can help you get the most out of them in a work context.
  • As tutor: AIs can offer personalized instruction and draw from a larger corpus than what any human can. That said, they can also hallucinate.
  • As coach: AIs can help us find patterns in data and help nudge us towards accomplishing our goals.

In all five cases, the emphasis is on collaborating with AI as a partner rather than a replacement. The book offers concrete examples of prompts and the subsequent responses. Some are about the writing of the book itself, a nice meta-touch.

The last chapter speculates about four possible futures, ranging from a scenario where the technology stalls to the opposite extreme where AI becomes all-powerful. The middle scenarios – slow growth and exponential growth – seem likeliest to me.

As I implied above, this is the best primer I’ve found on what most people currently understand by AI. The book is clear about the risks and challenges inherent in the technology, but also practical, informative, and philosophical. Much of its advice validates my experiences with the technology.

Co-Intelligence: Living and Working With AI