Ethan Mollick in Co-Intelligence

Humans, walking and talking bags of water and trace chemicals that we are, have managed to convince well-organized sand to pretend to think like us.

Why did I pick this book?

I was looking at books on artificial intelligence from a non-technical perspective. Most books currently on the topic of AI either are technical in nature (I am reading one such right now) or are focused on prompt engineering. This book came up in a podcast. The cover also caught my eye - the symbolism of AI as a forbidden fruit seemed apt to the topic.

If this book could be summarised into one word, it would be Catalyst.

If this book was a character, it would be a thoughtful mentor who can guide you through the possibilities of human-AI collaboration. This mentor is forward-thinking and possesses a sharp intellect. They acknowledge the potential pitfalls of over-reliance on AI while emphasising the importance of treating AI as a co-intelligence instead of an adversary. This partnership, the book suggests, holds the key to unlocking unprecedented levels of creativity and productivity.

What were the ideas that I liked and want to explore?

There were a few interesting ideas that I really liked and want to explore more:

• co-intelligence: The idea of using AI as a collaborator to achieve a human goal. The book explores how we can look at AIs in the roles of a person 🧑, a creative 🧑‍🎨, a coworker🧑‍💼, a tutor 🧑‍🏫, and a coach 🧢.
• The four co-intelligent principles:
  1. Always Invite AI to the Table.
  2. Be the Human in the Loop.
  3. Treat AI Like a Person (But Tell It What Kind of Person It Is).
  4. Assume This Is the Worst AI You Will Ever Use.
• The jagged frontier is the unpredictable and uneven nature of AI capabilities.
• The task classifications when using AI:
  • Just Me Tasks (no AI involvement)
  • Delegated Tasks (full AI involvement)
  • Centaur Tasks (humans and AI working in parallel)
  • Cyborg Tasks (humans and AI constantly collaborating on a single task).
• Using AI Personas for a better collaborative task solving.
• This trend of including AI prompts in books, as seen in Duly Noted and The Two But Rule, excites me.

How did I put this into action?

In the weeks after reading this book, I tried to put these principles into action and found them very useful.

• Reading Duly Noted further solidified the concept of co-intelligence. I used Google’s NotebookLM with my highlights and notes for this book as an input. I could clearly see the benefits as I used the tool to help me synthesis ideas from my notes.
• I’ve embraced the idea of using AI as an amanuensis firmly and I feel it’s an application of AI as a co-intelligence.

This book also piqued a few questions:

• [?] When a human working with an AI co-intelligence outperforms all but the best humans working without an AI, what does that mean for the ‘experts’?
• [?] With the improvements to the GPTs coming at break-neck speed, what’s the impact on human creativity vis-a-vis learning? Will we be engorging on AI-generated content more and more?
• [?] As LLMs improve, will ‘prompt engineering’ go away or get better with muti-modal inputs?
• [?] Is the Turing test really a test for humans not for machines?. If humans can be fooled into thinking that machines are sentient, then a Turing test taken by a machine is to test if the invigilator is human by trying to fool them.