Co-intelligence: Living and Working with AI by Prof. Ethan Mollick
The last 18 months have been awesomely crazy!
I have had my highs and lows. I stayed awake the whole night GPT4 was launched, admiring what OpenAI had created.
I have also found myself questioning the future of teaching “data science and Artificial Intelligence” to the world. What I thought was my calling has changed forever! I don’t completely understand how it will look in a few years.
In this fast-evolving, dynamic, and chaotic landscape, I came across the writings of Dr. Ethan Mollick (oneusefulthing.org). His writings have helped me think about this new world better than what I would have done without it.
So, when I came to know that he is writing a book on AI titled “Co-intelligence: Living and Working with AI,” I pre-ordered my copy and eagerly awaited the release. (Amazon link)
Who is Prof. Ethan Mollick?
Prof. Ethan Mollick teaches Entrepreneurship at The Wharton School. He is also the Faculty Director and Co-Founder of Wharton Interactive.
He was one of the first faculty to adapt chatGPT into his classroom and curriculum. For example – typical assignments in his class include creating an app, pitching to an AI-VC and many others.
All his classes are now AI mandatory, meaning you will need to face the new world if you attend any of them.
You can subscribe to his articles on Substack on oneusefulthing.org. I found all his writings fact-based, well-researched, practical, and simple to follow.
Who is this book for? What is it about?
This book is about living and working in the new world with AI. It keeps an individual (you) at the center and leaves you with a framework to navigate the new world by the end of the book.
Think of it as a tool to transform yourself from a Pre-chatGPT version of yourself to a Post-chatGPT version of yourself. There is no coding or technical knowledge required for you to read, learn and benefit from this book.
In a world where I get so many questions about the future of AI, its impact on the world, future jobs, privacy, deepfakes – the book answers a lot of questions about how to learn, live and survive this new world.
Prof. Mollick deliberately ignores the ethical aspects of AI and its implications to focus on what an individual can learn from using these AI tools.
The Four Rules of AI
The book lays out 4 principles about how to think and work with AI. Here are those 4 principles, with a very brief overview of them:
Principle 1: Always invite AI to the table.
One of the profound things Prof. Mollick says is to think about AI as an Alien we have to deal with. It is not obvious what tasks are done well with AI and what are not (he refers to this as a Jagged Frontier). The best way is to bring AI to every task you are doing and experience what it is good at. He recommends spending about 10 hours with any of the Frontier Models (best LLMs out there). Writing an article – take the help of AI; Creating a deck – take help; Coding … you get the drift! This can help you develop a nuanced understanding of what works with these AI tools.
Principle 2: Be the Human in the loop
AI can, at times, be a confident hallucinator, and unless you observe closely, the output can be full of errors and made-up things. Further, Prof. Mollick says the current state of AI is creating better-than-average human outcomes but not as good as an expert. So, you should map out what you are really good at, and you can benefit from working with AI on things you are not good at.
Principle 3 – Treat AI like a person
The third principle is that the current LLMs have been created to mimic human outcomes. They have been built on the vast human knowledge out there. So, to improve the quality of the outputs, treat and assign the role of a person to these AIs. Creating a marketing plan – ask AI to act as a CMO, working on a pitch deck – ask AI to act like a VC and give you the feedback. Just this single principle would enhance the quality of the outcomes you get from your prompts.
Principle 4 – The current AI models are the worst AIs we would see for the rest of our life
This is similar to what Sam Altman has said earlier. In a different interview, Prof. Mollick mentioned that every AI model you use today is already obsolete. The research labs are likely testing out the next version already.
I found these rules incredibly helpful to navigate and think about using AI in my work.
Different Roles of AI – A Person, A Creative, A Coworker, A Tutor and A Coach
The next part of the book details each of these roles that AI can take and their implications.
Given that my interests overlap with Prof. Mollick’s expertise (AI, Education, and Entrepreneurship), I found this incredibly useful. The possibility of having patient, personal tutors, and Coaches move me fundamentally.
3 profound impacts on the world of education, as laid out by Dr. Mollick, are:
- Homework as we know it is dead.
- AI will play a huge role in democratisation of education. It should build on the impact MOOCs have created in the world.
- New teaching pedagogies must be thought through. Dr. Mollick’s assignments are fun and challenging and give a glimpse of future pedagogies.
The Future of AI
Dr. Mollick lays out 4 possible scenarios about AI development from where we stand today:
- Scenario 1 – The current form of AI is the best AI we will see. He said this is the least likely scenario and I agree. Better models are coming out shortly.
- Scenario 2 – AI continues to evolve slowly / linearly (10-20% improvement) year on year. This makes life easy for humans; we can navigate a scenario like this and are used to it. But – I think the pace in the foreseeable future is going to be faster (Scenario 3 or 4)
- Scenario 3 – AI continues to evolve exponentially. We are definitely experiencing this. This leads to disruption across industries, and the world as we know it today will change quickly. The Flywheel effect of AI developing AI comes into the picture here.
- Scenario 4—The Machine God: We might end up with a machine with super-human intelligence or sentience. While multiple industry leaders are discussing this possibility, I still can’t visualize it completely.
Only time will tell which scenario will play out.
Other Thoughts / Learnings and Open Questions
While going through the book. I found some interesting thoughts I would probably build on in future. They did not come out in the summary above, so I am adding them to the article here.
Think of these as an interesting bucket list of ideas or open questions to think about.
- Prompt Engineering as a discipline is there only for the short term. In the next few iterations, the need and role of prompt engineers will go away to a large extent.
- I would have loved to hear more about using AI to create simulation environments and games. Dr. Mollick also leads Wharton Interactive, an initiative focused on building simulations for education. It looks like a topic worth a new book by itself.
- I would have also loved to hear more from Dr. Mollick on how we should think about applying Moore’s law for future development. This is slightly different from talking about AI evolution in the future (which he does!). We know computers have followed Moore’s law, but does he expect it to drive innovation across industries? Do we think we can apply some form of Moore’s law to AI capabilities? Would have loved to hear more on this from him.
- Another topic that Dr. Mollick touches on briefly is how we build future knowledge portals. A case in point is Stack Overflow. We could not have built current AIs if Stack Overflow did not exist. But in the current scenario, contributions to Stack Overflow have dropped. If we don’t spend time coding, will we still have coding experts in the future? What would serve as the source of improving intelligence for future LLM models? I guess—this is another article in itself.
Concluding Remarks
As you can see by now, I am extremely grateful to Dr. Mollick for what he has created, and I hope his writings are not going away any time soon.
What is clear is that we have entered a phase in which intelligence will evolve daily, and the world will look very different in a few years.
Do let me know your thoughts on the book and your takeaways from it.
Kunal is a post graduate from IIT Bombay in Aerospace Engineering. He has spent more than 10 years in field of Data Science. His work experience ranges from mature markets like UK to a developing market like India. During this period he has lead teams of various sizes and has worked on various tools like SAS, SPSS, Qlikview, R, Python and Matlab.
