Co-Intelligence

核心要点

• Treat AI as a co-worker to collaborate with, not a search engine or a calculator
• Invite AI into everything you do to learn its jagged frontier of strengths and weaknesses
• You must stay the human in the loop, checking, judging and owning the output
• Language models predict text, so they confidently make things up; verify anything that matters
• Skills, tasks and jobs will reshape around AI, so learn by experimenting now

本篇内容

1. A strange new colleague has arrived

   Imagine a new colleague joins your team. It has read more than any human alive, writes fluently in dozens of languages, never gets tired, and yet sometimes invents facts with total confidence.

2. Four rules for working with AI

   Mollick offers a handful of principles that shape the whole book. First, always invite AI to the table: try it on every task, even ones where you expect it to fail, because that is the only way to discover what it can really do.

3. The jagged frontier

   One of the book's most useful images is that AI capability is not a smooth line. It is a jagged frontier.

4. Centaurs and cyborgs

   How should the division of labour actually work? Mollick borrows two images from chess, where humans and computers have played together for years.

5. Why it makes things up

   To use AI well, you need a rough idea of what is inside. A language model is trained to predict the next word across an enormous amount of text.

6. Put it to work

   This week, pick 5 tasks you already do, from writing emails to preparing a lesson, and try each one with an AI tool, noting where it helped and where it stumbled. When you prompt, give the model a role, the context and the audience, then ask for several versions and choose or combine.