What do you think?
Rate this book
Co-Intelligence: Living and Working with AI
**A New York Times Bestseller**
'Co-Intelligence is the very best book I know about the ins, outs, and ethics of generative AI. Drop everything and read it cover to cover NOW' Angela Duckworth
Consumer AI has arrived. And with it, inescapable upheaval as we grapple with what it means for our jobs, lives and the future of humanity.
Cutting through the noise of AI evangelists and AI doom-mongers, Wharton professor Ethan Mollick has become one of the most prominent and provocative explainers of AI, focusing on the practical aspects of how these new tools for thought can transform our world. In Co-Intelligence, he urges us to engage with AI as co-worker, co-teacher and coach. Wide ranging, hugely thought-provoking and optimistic, Co-Intelligence reveals the promise and power of this new era.
'Co-Intelligence is the very best book I know about the ins, outs, and ethics of generative AI. Drop everything and read it cover to cover NOW' Angela Duckworth
Consumer AI has arrived. And with it, inescapable upheaval as we grapple with what it means for our jobs, lives and the future of humanity.
Cutting through the noise of AI evangelists and AI doom-mongers, Wharton professor Ethan Mollick has become one of the most prominent and provocative explainers of AI, focusing on the practical aspects of how these new tools for thought can transform our world. In Co-Intelligence, he urges us to engage with AI as co-worker, co-teacher and coach. Wide ranging, hugely thought-provoking and optimistic, Co-Intelligence reveals the promise and power of this new era.
243 pages, Kindle Edition
First published April 2, 2024
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 30 of 135 reviews
April 5, 2024
As someone who keeps up with Ethan Mollick's Substack and as someone who's quite interested in AI, I picked up this book instantly.
It's a solid layman's introduction to the current state of LLMs. Well-written, but if you've been playing around with AI yourself, there's likely nothing new here. This book is written for the rank beginner and does a good job at highlighting all the different ways LLMs can be used today, with examples.
Personally I was disappointed with how the book seemed content to give only a broad surface level overview of the state of LLMs. It would have been interesting for it to have dived deeper into the differences between Claude, GPT-4, Gemini etc, and it would've also been cool if it touched upon the implications of local models. However, I understand also that it's a rapidly changing space and devoting too much time to the nitty-gritty of what's happening today runs the risk of making the book obsolete in a few months. And to be fair, Ethan Mollick does in fact give his thoughts on the current frontier models on his Substack: https://open.substack.com/pub/oneusef...
All in all, it does what it does well. If I were to meet someone who is completely ignorant about LLMs and what can be done with LLMs today, I'd recommend this book.
For more detailed dives on AI, I'd recommend reading the articles on Mollick's Substack as well as keeping up with TheZvi's weekly AI roundup.
It's a solid layman's introduction to the current state of LLMs. Well-written, but if you've been playing around with AI yourself, there's likely nothing new here. This book is written for the rank beginner and does a good job at highlighting all the different ways LLMs can be used today, with examples.
Personally I was disappointed with how the book seemed content to give only a broad surface level overview of the state of LLMs. It would have been interesting for it to have dived deeper into the differences between Claude, GPT-4, Gemini etc, and it would've also been cool if it touched upon the implications of local models. However, I understand also that it's a rapidly changing space and devoting too much time to the nitty-gritty of what's happening today runs the risk of making the book obsolete in a few months. And to be fair, Ethan Mollick does in fact give his thoughts on the current frontier models on his Substack: https://open.substack.com/pub/oneusef...
All in all, it does what it does well. If I were to meet someone who is completely ignorant about LLMs and what can be done with LLMs today, I'd recommend this book.
For more detailed dives on AI, I'd recommend reading the articles on Mollick's Substack as well as keeping up with TheZvi's weekly AI roundup.
May 13, 2024
Clear-eyed primer to artificial intelligence, specifically the Large Language Models and generative AI now broadly available. I first heard Mollick on this episode of Ezra Klein's podcast, and appreciated his structured suggestions for using AI in daily tasks for work and general living.
Many of these same principles are repeated and expounded on in Chapter 3 of CO-INTELLIGENCE:
-- Invite AI to the table - innovation - using AI as a tool, not a crutch
--Learn to be the human in the loop -"There is no there there, despite it sounding like a human." Check the AI for lies, misinformation, and inaccuracy. Sharpen your own general and subject-matter knowledge and critical thinking, making you responsible for the output that you get from the AI
--Treat the AI like a person, but tell it what kind of person to be - conversational queries and tone yields the best results. Anthropomorphism is natural to the human mind, but the more we ascribe false agency to AI, the more we are exploited.
--This is the worst AI you will ever use - the models are constantly changing, as are you, so the results you get now will therefore be less "good" than anything you get in the future.
Mollick details the creative and collaborative ways that AI is being used, and cautiously reminds of the ways that these tools can be used for ill, encouraging a cautious but exploratory approach. There is not an all-out bowing to the overlords, or the opposite "the sky is falling"/hair on fire tone here - neutral and clear, this was a useful read for this moment in time in the evolutions of LLMs.
Audiobook note: Mollick reads his own words here, and when he is sharing AI-generated and polished text, an AI-generated voice reads it on the audiobook. Authentic and honest touch.
Many of these same principles are repeated and expounded on in Chapter 3 of CO-INTELLIGENCE:
-- Invite AI to the table - innovation - using AI as a tool, not a crutch
--Learn to be the human in the loop -"There is no there there, despite it sounding like a human." Check the AI for lies, misinformation, and inaccuracy. Sharpen your own general and subject-matter knowledge and critical thinking, making you responsible for the output that you get from the AI
--Treat the AI like a person, but tell it what kind of person to be - conversational queries and tone yields the best results. Anthropomorphism is natural to the human mind, but the more we ascribe false agency to AI, the more we are exploited.
--This is the worst AI you will ever use - the models are constantly changing, as are you, so the results you get now will therefore be less "good" than anything you get in the future.
Mollick details the creative and collaborative ways that AI is being used, and cautiously reminds of the ways that these tools can be used for ill, encouraging a cautious but exploratory approach. There is not an all-out bowing to the overlords, or the opposite "the sky is falling"/hair on fire tone here - neutral and clear, this was a useful read for this moment in time in the evolutions of LLMs.
Audiobook note: Mollick reads his own words here, and when he is sharing AI-generated and polished text, an AI-generated voice reads it on the audiobook. Authentic and honest touch.
April 8, 2024
Really informative for those new to AI or LLMs and want a good overview of the impact it will have on our lives. For those already working in the AI space, although this does have some best practices - you won't find a ton of new insights here.
It also vaguely touched on the ethics of training models and politics but it mainly focused on the following topics:
- Understanding the limitations of AI
- Best practices for using AI for creative tasks & idea generation
- Approaches to human and AI collaboration
- Productivity and positive impacts of AI
- Impacts on education, workforce and innovation
A very timely and interesting read but I imagine most of this information will feel out of date in a year , so I'd pick this up soon if you're interested.
It also vaguely touched on the ethics of training models and politics but it mainly focused on the following topics:
- Understanding the limitations of AI
- Best practices for using AI for creative tasks & idea generation
- Approaches to human and AI collaboration
- Productivity and positive impacts of AI
- Impacts on education, workforce and innovation
A very timely and interesting read but I imagine most of this information will feel out of date in a year , so I'd pick this up soon if you're interested.
April 8, 2024
Perhaps not as impactful a text as I’d hoped but thoughtful and pragmatically optimistic. I didn’t learn much that was new but I picked up a lot of useful thoughts to share with the more AI-phobic of my friends and colleagues.
May 19, 2024
Humans, walking and talking bags of water and trace chemicals that we are, have managed to convinced well organised sand to pretend to think like us.
"Co-Intelligence: Living and Working with AI" is an easily digestible and thought-provoking text on the current state of play with Generative AI written by Ethan Mollick Wharton School of the University of Pennsylvania. 'Current' a key word and I suspect once this book is quickly read, the reader would be best following his Substack One Useful Thing for ongoing developments, including this entry where he describes how he used AI in writing this book.
He gives four principles for how we, as individuals, should interact with AI:
Principle 1: Always invite AI to the table
Principle 2: Be the human in the loop
Principle 3: Treat AI like a person (but tell it what kind of person it is)
Principle 4: Assume this is the worst AI you will ever use
Expanding on the first he says, in advice that present a challenge to companies for how to best tap their resources (a challenge he later addresses):
You should try inviting Al to help you in everything you do, barring legal or ethical barriers. As you experiment, you may find that AI help can be satisfying, or frustrating, or useless, or unnerving. But you aren't just doing this for help alone; familiarizing yourself with AI's capabilities allows you to better un-derstand how it can assist you—or threaten you and your job. Given that AI is a General Purpose Technology, there is no single manual or instruction book that you can refer to in order to understand its value and its limits.
...
And this experimentation gives you the chance to become the best expert in the world in using AI for a task you know well. The reason for this stems from a fundamental truth about innovation: it is expensive for organizations and companies but cheap for individuals doing their job. Innovation comes from trial and error, which means that an organization trying to launch a new product to help a marketer write more compelling copy would need to build the product, test it on many users, and make changes many times to make something that works. A marketer, however, is writing copy all the time and can experiment with many different ways of using AI to help until they find one that succeeds. No need to hire a team or use expensive software development cycles.
The blurb describes the book as optimistic, although I'd say instead Mollick puts a 'make the best of the situation' spin on where we now are and likely futures - the fact that the cover picture is from van der Goes 'The Fall', with Adam about to pluck the apple from the tree, is rather indicative of what the book portrays, and, even if not intended, it left me feeling deeply uneasy.
I can't really let this pass without asking AI to write a review for me, so this is Open AI's attempt (ChatGPT4.o) using information it could gather from the intranet. I didn't ask it to actually read the book:
"Co-Intelligence: Living and Working with AI" is an easily digestible and thought-provoking text on the current state of play with Generative AI written by Ethan Mollick Wharton School of the University of Pennsylvania. 'Current' a key word and I suspect once this book is quickly read, the reader would be best following his Substack One Useful Thing for ongoing developments, including this entry where he describes how he used AI in writing this book.
He gives four principles for how we, as individuals, should interact with AI:
Principle 1: Always invite AI to the table
Principle 2: Be the human in the loop
Principle 3: Treat AI like a person (but tell it what kind of person it is)
Principle 4: Assume this is the worst AI you will ever use
Expanding on the first he says, in advice that present a challenge to companies for how to best tap their resources (a challenge he later addresses):
You should try inviting Al to help you in everything you do, barring legal or ethical barriers. As you experiment, you may find that AI help can be satisfying, or frustrating, or useless, or unnerving. But you aren't just doing this for help alone; familiarizing yourself with AI's capabilities allows you to better un-derstand how it can assist you—or threaten you and your job. Given that AI is a General Purpose Technology, there is no single manual or instruction book that you can refer to in order to understand its value and its limits.
...
And this experimentation gives you the chance to become the best expert in the world in using AI for a task you know well. The reason for this stems from a fundamental truth about innovation: it is expensive for organizations and companies but cheap for individuals doing their job. Innovation comes from trial and error, which means that an organization trying to launch a new product to help a marketer write more compelling copy would need to build the product, test it on many users, and make changes many times to make something that works. A marketer, however, is writing copy all the time and can experiment with many different ways of using AI to help until they find one that succeeds. No need to hire a team or use expensive software development cycles.
The blurb describes the book as optimistic, although I'd say instead Mollick puts a 'make the best of the situation' spin on where we now are and likely futures - the fact that the cover picture is from van der Goes 'The Fall', with Adam about to pluck the apple from the tree, is rather indicative of what the book portrays, and, even if not intended, it left me feeling deeply uneasy.
I can't really let this pass without asking AI to write a review for me, so this is Open AI's attempt (ChatGPT4.o) using information it could gather from the intranet. I didn't ask it to actually read the book:
Ethan Mollick's "Co-Intelligence: Living and Working with AI" is a timely exploration of how artificial intelligence is transforming various aspects of human life. Drawing from his experience as a professor at Wharton and the insights shared through his "One Useful Thing" newsletter, Mollick presents a compelling case for viewing AI as a partner in our professional and personal endeavors.
The book's strength lies in its practical approach. Mollick vividly illustrates how AI can enhance our capabilities, using examples from business and education. He describes scenarios where AI tools, such as ChatGPT, can undertake complex tasks like teaching negotiation skills, which traditionally required extensive human input. This hands-on perspective is both engaging and informative, making complex AI concepts accessible to a broad audience.
Mollick’s optimism about AI is balanced by a thoughtful consideration of its ethical implications and the need to preserve human identity in an increasingly automated world. He encourages readers to actively engage with AI, harnessing its potential to create a better future while remaining vigilant about its challenges.
However, a notable concern with "Co-Intelligence" is its potential to become outdated quickly. AI is a rapidly evolving field, with new developments emerging almost daily. While Mollick provides a thorough analysis of current AI capabilities and applications, the relentless pace of technological advancement means that some insights may soon be superseded by new innovations. Readers will need to stay updated with the latest trends to fully benefit from the book’s teachings.
Despite this, "Co-Intelligence" remains a valuable resource. Mollick’s lucid writing and practical examples offer a clear roadmap for integrating AI into our lives. For anyone looking to understand the transformative power of AI and how to work alongside these new technologies, Mollick's book is an essential read, even if it requires future updates to remain fully relevant.
May 2, 2024
I used to be wary that a lot of tech demos were faked (and some of them still are) but now with chat that anyone can use on their own device, one can prove that it does indeed work, which is amazing - but the issue now seems to be that users convince themselves the product they are using is something it is not.
Here is an author who has moved briskly to publish yet another book that has GPT-4 copypastes in it, which one can always completely skip, and with it a lot of unsubstantiated, imprecise allusions, skimping on the details but namechecking peers. For some reason endnotes are not included in the text.
The whole framing of the alien mind is lazy and inconsistent with what we know about how these models are developed which only highlights the gullibility of the author, who seems to be stuck in a reasoning trap that presupposes a/ that it works perfectly now, which it doesn't and b/things are going to happen exactly according to his assumptions, not least of which the idea of unbounded improvement or the supremacy of chat as a modality which to me already both feel no longer true.
The book has a tendency to hand wave all deeper exploration away to other experts who presumably lag in the darkness somewhere. I guess if you were completely non-technical or had not been in grad school in 20 years this book might be interesting to catch up on what’s going on, but given how shallow it is I can’t recommend it, and at the end I got bored of entertaining the author’s AGI fantasies.
Here is an author who has moved briskly to publish yet another book that has GPT-4 copypastes in it, which one can always completely skip, and with it a lot of unsubstantiated, imprecise allusions, skimping on the details but namechecking peers. For some reason endnotes are not included in the text.
The whole framing of the alien mind is lazy and inconsistent with what we know about how these models are developed which only highlights the gullibility of the author, who seems to be stuck in a reasoning trap that presupposes a/ that it works perfectly now, which it doesn't and b/things are going to happen exactly according to his assumptions, not least of which the idea of unbounded improvement or the supremacy of chat as a modality which to me already both feel no longer true.
The book has a tendency to hand wave all deeper exploration away to other experts who presumably lag in the darkness somewhere. I guess if you were completely non-technical or had not been in grad school in 20 years this book might be interesting to catch up on what’s going on, but given how shallow it is I can’t recommend it, and at the end I got bored of entertaining the author’s AGI fantasies.
April 19, 2024
Very short interesting introduction for people not very familiar with Artificial Intelligence (AI) and its potential uses. It's not very in depth though.
As a result, I asked ChatGPT: What are the best books about AI?
ChatGPT: There are many excellent books about AI that cover various aspects of the field, from its history and development to its potential impact on society. Here are some highly regarded ones:
1. "Life 3.0: Being Human in the Age of Artificial Intelligence" by Max Tegmark - Explores the societal implications of AI and its potential to shape the future of humanity.
2. "Superintelligence: Paths, Dangers, Strategies" by Nick Bostrom - Examines the potential risks and benefits of artificial superintelligence and how it might impact humanity.
3. "AI Superpowers: China, Silicon Valley, and the New World Order" by Kai-Fu Lee - Provides insights into the AI landscape, exploring the competition between China and Silicon Valley and its implications for the future.
4. "The Master Algorithm: How the Quest for the Ultimate Learning Machine Will Remake Our World" by Pedro Domingos - Explores the concept of a master algorithm that could unlock the secrets of learning and reshape society.
5. "Artificial Unintelligence: How Computers Misunderstand the World" by Meredith Broussard - Discusses the limitations and misunderstandings of AI and the importance of human intelligence in the age of automation.
6. "AIQ: How People and Machines Are Smarter Together" by Nick Polson and James Scott - Explores the intersection of human and artificial intelligence and how they can work together to solve complex problems.
7. "Human Compatible: Artificial Intelligence and the Problem of Control" by Stuart Russell - Examines the challenge of ensuring that AI systems align with human values and goals to prevent unintended consequences.
8. "Machine Learning: A Probabilistic Perspective" by Kevin P. Murphy - Provides a comprehensive introduction to machine learning from a probabilistic viewpoint, suitable for both beginners and experts.
9. "Deep Learning" by Ian Goodfellow, Yoshua Bengio, and Aaron Courville - Offers a thorough overview of deep learning techniques, algorithms, and applications.
10. "Artificial Intelligence: A Guide for Thinking Humans" by Melanie Mitchell - Explores fundamental concepts of AI in an accessible manner, helping readers understand its potential and limitations.
These books cover a wide range of topics within the field of AI and are highly recommended for anyone interested in learning more about this rapidly evolving technology.
As a result, I asked ChatGPT: What are the best books about AI?
ChatGPT: There are many excellent books about AI that cover various aspects of the field, from its history and development to its potential impact on society. Here are some highly regarded ones:
1. "Life 3.0: Being Human in the Age of Artificial Intelligence" by Max Tegmark - Explores the societal implications of AI and its potential to shape the future of humanity.
2. "Superintelligence: Paths, Dangers, Strategies" by Nick Bostrom - Examines the potential risks and benefits of artificial superintelligence and how it might impact humanity.
3. "AI Superpowers: China, Silicon Valley, and the New World Order" by Kai-Fu Lee - Provides insights into the AI landscape, exploring the competition between China and Silicon Valley and its implications for the future.
4. "The Master Algorithm: How the Quest for the Ultimate Learning Machine Will Remake Our World" by Pedro Domingos - Explores the concept of a master algorithm that could unlock the secrets of learning and reshape society.
5. "Artificial Unintelligence: How Computers Misunderstand the World" by Meredith Broussard - Discusses the limitations and misunderstandings of AI and the importance of human intelligence in the age of automation.
6. "AIQ: How People and Machines Are Smarter Together" by Nick Polson and James Scott - Explores the intersection of human and artificial intelligence and how they can work together to solve complex problems.
7. "Human Compatible: Artificial Intelligence and the Problem of Control" by Stuart Russell - Examines the challenge of ensuring that AI systems align with human values and goals to prevent unintended consequences.
8. "Machine Learning: A Probabilistic Perspective" by Kevin P. Murphy - Provides a comprehensive introduction to machine learning from a probabilistic viewpoint, suitable for both beginners and experts.
9. "Deep Learning" by Ian Goodfellow, Yoshua Bengio, and Aaron Courville - Offers a thorough overview of deep learning techniques, algorithms, and applications.
10. "Artificial Intelligence: A Guide for Thinking Humans" by Melanie Mitchell - Explores fundamental concepts of AI in an accessible manner, helping readers understand its potential and limitations.
These books cover a wide range of topics within the field of AI and are highly recommended for anyone interested in learning more about this rapidly evolving technology.
Want to read
March 25, 2024Recommended by Adam Grant
https://adamgrant.substack.com/p/the-...
Podcast: ChatGPT did NOT title this podcast (w/ Allie Miller & Ethan Mollick)
https://podcasts.apple.com/us/podcast...
https://adamgrant.substack.com/p/the-...
Podcast: ChatGPT did NOT title this podcast (w/ Allie Miller & Ethan Mollick)
https://podcasts.apple.com/us/podcast...
May 5, 2024
When ChatGPT3 was first launched in late 2022, I was teaching Albert Camus' The Stranger and I was immediately worried about how it might impact the curriculum and the assignments for my class. Experimenting with ChatGPT and designing different prompts, I was not at first impressed with the quality of its answers (I thought its grasp of the text was vague and its prose stilted and lacking panache) but I was soon awed by its versatile ability to respond creatively to novel questions. In one case, I asked it to compare The Stranger and Where's Waldo?. Naively, I had assumed that this comparative exercise, involving two vastly different genres of books, one a philosophical novel, the other a children's puzzle-book, would stump the computer. But it immediately returned a plausible essay arguing that whereas The Stranger depicts a protagonist who is always an outsider in his community, Where's Waldo? centered around a character who always blends in and hides in any environment. The discussion of both texts was cursory but the thesis was surprisingly creative and perceptive.
Ethan Mollick's recent book is an exploration of the "jagged frontier" of Large Language Models. As I discovered in my first forays into this technology, it is immensely difficult to predict the exact contours of AI competency. ChatGPT can write sophisticated essays on an array of topics. It can compose rhyming sonnets. It can even translate languages. But at the same time, it might invent quotations from a book and fail at basic language arts (such as writing the word "hello" backwards or finding an anagram of "stop"). It might be able to give you the solutions to an advanced mathematical problem and yet make elementary mistakes in simple arithmetic. If you ask it for a random number between 1 and 100, it will frequently return the number of 42 (presumably because this is a common number on the internet, an allusion to Hitchhiker's Guide to the Galaxy). It will successfully write the code for a game of tic-tac-toe and yet be unable to play the game itself. Similarly, it can spell out the rules of chess but will often make novice mistakes and illegal moves if you challenge it to a match. As Ethan Mollick's book makes clear, AI is often good at creative work and at finding unexpected connections, but it can still flunk the most basic intellectual tasks. At its core, ChatGPT is a statistical model of language, producing a plausible imitation of human responses by finely calculating the probability of clusters of words—there is no cognition, no knowledge, no logical validation underlying its answers, nothing but huge stochastic calculations of what words should go next to other words.
What Ethan Mollick's book offers is not a comprehensive guide to AI or an evaluation of different LLMs. Instead, his book gives guiding principles for the practical and ethical use of AI. In four parts, he articulates a general framework for interacting with AI: first, he exhorts readers to use AI, to explore and map the frontiers of its capabilities, always seeing how AI handles both the mundane and the higher-order tasks of every job. Second, he reminds readers that they always need to be active workers rather than passive consumers, diligently checking AI responses for inaccuracies and hallucinations. Third, he argues that users should treat AI as a person, inviting it to respond to prompts in different personas, critiquing its answers and reformulating questions. Just as you would ask questions, clarify where necessary, and offer feedback to a colleague, so too should you extend this grace to AI. Curiously, Google's most advanced LLM was found to be more effective when prompts contained the words, "Take a deep breath and work on this problem step by step"—not because the AI was hyperventilating or experiencing anxiety but because something about the wording elicited a slightly different and more accurate language model. So interact with AI as a person, for whom variable instructions will result in variable answers. Finally, he argues that every user needs to treat every LLM as if it is the worst AI they "will ever use", recognizing that the next iteration or the next model might be superior.
There is a lot of uncertainty about the future of AI, work and education and, in some quarters, even apocalyptic fantasies of some computerized superintelligence imminently taking dominion over the world. Reading Mollick, however, I am struck by the banality of AI. As more and more workers begin to use AI to draft emails, to summarize articles, to edit reports, to brainstorm ideas and slogans, language will simply become less special. AI will be the presumptive co-author of every written text and the conventional stuff of an email (the salutations, the professional tone, the warmth of expression) will simply be interpreted as automated formulas. We are heading into a future in which AI machines will write cover letters for job-seekers which other AI machines will interpret, sort and rank for employers. College professors will use AI to write elegant and persuasive letters for their many students, and readers will assume that AI underpins the words they are reading. An employer will one day receive a note of thanks or a performance review wholly generated by AI without any human supervision. Mollick describes this future as a "crisis of meaning" but to put it more specifically, language models will result in a crisis of language: no one will know if the words they are reading were written by conscious minds that think and feel the sentiments expressed. Anything online will be as generic and impersonal as a Hallmark card.
So while AI will optimize many "knowledge work" occupations, it will ultimately banalize language. The well-written essay will be as quotidian as microwave pizza and assembly-line cars; the polite email will be an engineered datum. As a result, humans will need to find new areas of value. Will AI be able to produce witty jokes that satirize the current moment? Will AI be able to write introspective poems that capture the human experience? Humans will have to discover new thresholds for creativity. Mollick's book is a useful book for schools and businesses but the big questions are left in the air.
Ethan Mollick's recent book is an exploration of the "jagged frontier" of Large Language Models. As I discovered in my first forays into this technology, it is immensely difficult to predict the exact contours of AI competency. ChatGPT can write sophisticated essays on an array of topics. It can compose rhyming sonnets. It can even translate languages. But at the same time, it might invent quotations from a book and fail at basic language arts (such as writing the word "hello" backwards or finding an anagram of "stop"). It might be able to give you the solutions to an advanced mathematical problem and yet make elementary mistakes in simple arithmetic. If you ask it for a random number between 1 and 100, it will frequently return the number of 42 (presumably because this is a common number on the internet, an allusion to Hitchhiker's Guide to the Galaxy). It will successfully write the code for a game of tic-tac-toe and yet be unable to play the game itself. Similarly, it can spell out the rules of chess but will often make novice mistakes and illegal moves if you challenge it to a match. As Ethan Mollick's book makes clear, AI is often good at creative work and at finding unexpected connections, but it can still flunk the most basic intellectual tasks. At its core, ChatGPT is a statistical model of language, producing a plausible imitation of human responses by finely calculating the probability of clusters of words—there is no cognition, no knowledge, no logical validation underlying its answers, nothing but huge stochastic calculations of what words should go next to other words.
What Ethan Mollick's book offers is not a comprehensive guide to AI or an evaluation of different LLMs. Instead, his book gives guiding principles for the practical and ethical use of AI. In four parts, he articulates a general framework for interacting with AI: first, he exhorts readers to use AI, to explore and map the frontiers of its capabilities, always seeing how AI handles both the mundane and the higher-order tasks of every job. Second, he reminds readers that they always need to be active workers rather than passive consumers, diligently checking AI responses for inaccuracies and hallucinations. Third, he argues that users should treat AI as a person, inviting it to respond to prompts in different personas, critiquing its answers and reformulating questions. Just as you would ask questions, clarify where necessary, and offer feedback to a colleague, so too should you extend this grace to AI. Curiously, Google's most advanced LLM was found to be more effective when prompts contained the words, "Take a deep breath and work on this problem step by step"—not because the AI was hyperventilating or experiencing anxiety but because something about the wording elicited a slightly different and more accurate language model. So interact with AI as a person, for whom variable instructions will result in variable answers. Finally, he argues that every user needs to treat every LLM as if it is the worst AI they "will ever use", recognizing that the next iteration or the next model might be superior.
There is a lot of uncertainty about the future of AI, work and education and, in some quarters, even apocalyptic fantasies of some computerized superintelligence imminently taking dominion over the world. Reading Mollick, however, I am struck by the banality of AI. As more and more workers begin to use AI to draft emails, to summarize articles, to edit reports, to brainstorm ideas and slogans, language will simply become less special. AI will be the presumptive co-author of every written text and the conventional stuff of an email (the salutations, the professional tone, the warmth of expression) will simply be interpreted as automated formulas. We are heading into a future in which AI machines will write cover letters for job-seekers which other AI machines will interpret, sort and rank for employers. College professors will use AI to write elegant and persuasive letters for their many students, and readers will assume that AI underpins the words they are reading. An employer will one day receive a note of thanks or a performance review wholly generated by AI without any human supervision. Mollick describes this future as a "crisis of meaning" but to put it more specifically, language models will result in a crisis of language: no one will know if the words they are reading were written by conscious minds that think and feel the sentiments expressed. Anything online will be as generic and impersonal as a Hallmark card.
So while AI will optimize many "knowledge work" occupations, it will ultimately banalize language. The well-written essay will be as quotidian as microwave pizza and assembly-line cars; the polite email will be an engineered datum. As a result, humans will need to find new areas of value. Will AI be able to produce witty jokes that satirize the current moment? Will AI be able to write introspective poems that capture the human experience? Humans will have to discover new thresholds for creativity. Mollick's book is a useful book for schools and businesses but the big questions are left in the air.
April 17, 2024
As artificial intelligence becomes increasingly advanced, Ethan Mollick argues in his book that we are entering an era of "co-intelligence" - where humans and AI systems will work together in partnership.
The central idea is that instead of viewing AI as just a tool, we need to treat it more like a co-worker or collaborator to get the best out of this transformative technology across all industries and aspects of life.
Mollick provides real examples and strategies for taking advantage of AI's capabilities while ensuring human ingenuity, ethics and identity remain central. He guides readers on how to optimistically embrace AI, while continuously adapting and improving upon it.
In simple terms, Co-Intelligence lays out a vision for the future where humans and intelligent machines cooperate in a mutually beneficial way, instead of an "us vs them" mindset. It's a thought-provoking read to prepare for the AI revolution.
The central idea is that instead of viewing AI as just a tool, we need to treat it more like a co-worker or collaborator to get the best out of this transformative technology across all industries and aspects of life.
Mollick provides real examples and strategies for taking advantage of AI's capabilities while ensuring human ingenuity, ethics and identity remain central. He guides readers on how to optimistically embrace AI, while continuously adapting and improving upon it.
In simple terms, Co-Intelligence lays out a vision for the future where humans and intelligent machines cooperate in a mutually beneficial way, instead of an "us vs them" mindset. It's a thought-provoking read to prepare for the AI revolution.
May 10, 2024
This is a good clear summary of where the technology for large language models stands today. Also a good summary of how to think about these tools. I emphasize the word today as this technology progresses over the next few years, parts of this book will become obsolete or out of date.
April 14, 2024
The main takeaway for me is how effectively we can use prompts to utilize large language models (LLMs). It has practical value immediately.
April 7, 2024
“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.”
I love Ethan Mollick’s newsletter One Useful Thing and this book did not disappoint. Great book for how to think about generative AI, not lists of prompt writing. And humorous throughout.
I love Ethan Mollick’s newsletter One Useful Thing and this book did not disappoint. Great book for how to think about generative AI, not lists of prompt writing. And humorous throughout.
May 19, 2024
Combine this book with Ethan Mollick's Substack for great practical insights into the latest AI technologies.
Co-Intelligence is a pretty balanced and comprehensive approach to applying generative AI tools to daily life tasks. He explores how to leverage genAI as a creative, coworker, tutor, and coach. For those familiar with the latest LLMs like GTP4, Claude, or Gemini, then a lot of this material will be familiar. It may even been too superficial in some ways (which is why I mention going to his Substack), but it is good coverage that everyone needs to know. It is delivered in an easily digestible way too.
Co-Intelligence is a pretty balanced and comprehensive approach to applying generative AI tools to daily life tasks. He explores how to leverage genAI as a creative, coworker, tutor, and coach. For those familiar with the latest LLMs like GTP4, Claude, or Gemini, then a lot of this material will be familiar. It may even been too superficial in some ways (which is why I mention going to his Substack), but it is good coverage that everyone needs to know. It is delivered in an easily digestible way too.
April 30, 2024
I listened to this on audiobook & it was the first time I had to listen to a book BELOW normal speed. I listened on .95 instead of my typical 1.25, which was partially because of the narrator and partially because of the content being so much to process.
Overall, it was a good read. I am wary yet curious about AI and appreciated some of the information outlined in this. The author obviously has significant experience interacting with LLMs/chatbots and it was nice to hear him outline the benefits and limitations of them. I would've enjoyed going a little bit deeper on future and current implications, and perhaps even exploring more of non-LLM AI. Still, I liked it and I feel like I can think more evenly and openly about AI after reading this.
Overall, it was a good read. I am wary yet curious about AI and appreciated some of the information outlined in this. The author obviously has significant experience interacting with LLMs/chatbots and it was nice to hear him outline the benefits and limitations of them. I would've enjoyed going a little bit deeper on future and current implications, and perhaps even exploring more of non-LLM AI. Still, I liked it and I feel like I can think more evenly and openly about AI after reading this.
April 27, 2024
Llms are just giant copying machines. Excellent…innovation is all about copying. lots of ways to tame chat GPT and put it to work. Must read for folks entering the workforce or interested in how to use the tech, and avoid falling victim to its inevitable rise and domination…
May 2, 2024
Really interesting near-term look at AI for novices (me) on the topic. Some parts had me actually saying “holy cow” out loud as I was reading them.
April 23, 2024
Practical and approachable perspective on how to think about generative AI and its potential impacts to how we learn and work. Fans of the newsletter will find some of the content repeated but well curated into a consistent narrative.
May 3, 2024
The opposite of timeless. Builds on hype, guesswork and current state, lacking any depth that would make it stay relevant beyond the next few months.
If you wish to understand the real potential and drawbacks of AI solutions, go for “The Alignment Problem” by Brian Christian.
If you wish to understand the real potential and drawbacks of AI solutions, go for “The Alignment Problem” by Brian Christian.
April 16, 2024
This is a great primer on artificial intelligence. The 5 stars are mostly because the topic is so current. The book offers some theory and a lot of practical tips for how to start using AI, how to think of AI and what it could become. If you're interested in AI -- and we all might want to be right now -- read it.
April 15, 2024
This book is for now Q2 2024
Please read it; it will be a quick read and will open your eyes to how quickly things are changing and how to start using AI in your daily life. The part of education for children and teens is exciting. I may make some Shane for my kid's education path.
Please read it; it will be a quick read and will open your eyes to how quickly things are changing and how to start using AI in your daily life. The part of education for children and teens is exciting. I may make some Shane for my kid's education path.
April 9, 2024
3.5 stars. This was a good overview of the current state of AI (as of late 2023). It discusses what AI is good at and not good at, and how one might approach leveraging AI in different areas given these strengths and weaknesses. It doesn’t go into any in-depth detail about prompt engineering (and suggests that the idea of prompt engineering itself probably has a limited lifetime), but discusses more generally how to get more interesting and relevant results from AI interactions. The book also discusses the likely impact of AI on different areas of society and human endeavor. Finally, it lays out a vision of different possibilities of where an AI focused future might take us. Overall I felt like the author payed lip service to some AI concerns, but seemed to quickly move past those with an “all that said…” and into discussions of advantages and opportunities.
April 25, 2024
Strikingly unremarkable. If you’ve paid even passing attention to AI recently, there is nothing here to surprise you. If you told me the entire thing was written by AI, I wouldn’t bat an eye. It’s basically a list of neat things AI can do, written without style. And some of the main points feel oddly and unresolvedly contradictory, e.g., the idea that education will become less valuable in the workforce because of AI vs. the gushing about how AI can be used to democratize education.
And then it ends with a weirdly apocalyptic set of possible scenarios the world appears headed for. Apropos of none of the neato things discussed in the previous 98% of the book.
And then it ends with a weirdly apocalyptic set of possible scenarios the world appears headed for. Apropos of none of the neato things discussed in the previous 98% of the book.
April 10, 2024
I had read a couple of Ethan Mollick’s articles through Substack and got the book on Kindle after listening to his interview with Ezra Klein. https://www.nytimes.com/2024/04/02/op...
It’s a great summary, specifically on generative AI and LLM’s. Pragmatic take on how we can use these tools for the better
It’s a great summary, specifically on generative AI and LLM’s. Pragmatic take on how we can use these tools for the better
May 3, 2024
I never say this but get the audiobook instead of the hard copy because you need to hear a chat bot read the recipe for napalm in a pirates’ voice. Thank me later.
April 7, 2024
With the rate of change in AI, I imagine it can be quite difficult to write a book and have it still be poignant and relevant upon release, let alone as the years pass.
Ethan Mollick has done an admirable job of achieving this feat.
AI is relevant in everyone’s life and will be increasingly so as technology advances. This is a very accessible introduction to AI, replete with enough insight and contemplation on its future to make it a worthwhile read for most.
Ethan Mollick has done an admirable job of achieving this feat.
AI is relevant in everyone’s life and will be increasingly so as technology advances. This is a very accessible introduction to AI, replete with enough insight and contemplation on its future to make it a worthwhile read for most.
April 6, 2024
A great read for anyone even remotely interested in AI
A great read for anyone even remotely interested in (or worried about) AI. Balanced, well-researched, and with practical insights beginners can test for themselves. Written in an accessible way; enjoyable reading for a thought-provoking topic.
A great read for anyone even remotely interested in (or worried about) AI. Balanced, well-researched, and with practical insights beginners can test for themselves. Written in an accessible way; enjoyable reading for a thought-provoking topic.
April 29, 2024
Straightforward thinking about how AI will impact us from an expert. Good writing and good points. I follow the author on social media, so I’ve heard him make many of these points already, but I don’t regret reading the full book.
Shelved as 'partial'
April 27, 2024Skimmed through this. Very surface level, LLMs 101. Not worth it; there’s plenty out there on the topic that’s better than this. The strategies for “using” AI are barebones and pedantic.
April 27, 2024
Wharton Professor with deep ties to Artificial Intelligence (i.e., Marvin Minsky) produces a "how-to" guide for incorporating LLMs into life and business. Ethan Mollick's blend of personal insights—like using GPT to reframe personal setbacks—and broader technological implications offers a fresh, breezy narrative that is educational and thought-provoking.
Consistent with my standard learning practice, I listened to many podcasts where Mollick introduced and contextualized the book. Worth noting that Professor Mollick now REQUIRES students to use AI in his entrepreneurship course to enable students to "do the impossible" (i.e., build an app).
The Power of "Co-Intelligence": Humans and AI Working Together:
Central to the book's theme is "co-intelligence," which emphasizes the collaborative relationship between humans and AI systems. Mollick argues that the most successful organizations will be those that harness the power of AI to augment and enhance human capabilities rather than simply replace human workers. For example, AI can automate routine tasks, allowing humans to focus on higher-level, creative work that requires critical thinking and problem-solving skills.
We are all "User Innovators":
Mollick, who worked with AI pioneer Marvin Minsky, states, "We're all experiencing this ride into a new AI wave and we're all figuring out this 'general purpose technology' together." In real time, we are defining its utility and use cases.
The Unpredictable Nature of AI:
Mollick highlights the unpredictable and unreliable nature of AI, noting that "it can surprise us with novel solutions, forget its own abilities, and hallucinate incorrect answers." He emphasizes the need for experimentation and learning when incorporating AI into processes, as there is no definitive guide on how to use it in organizations.
The Importance of Human Expertise:
While AI can assist in various tasks, Mollick stresses the importance of human expertise in evaluating AI-generated outputs. He provides examples such as a seasoned architect assessing an AI-proposed building plan or a skilled physician scrutinizing an AI-generated diagnosis. Mollick argues that building expertise requires a strong knowledge base, deliberate practice, feedback, and structured reflection.
The Equalizing Potential of AI:
Mollick emphasizes AI's potential to democratize access to powerful tools on a global scale. He notes that GPT-4, a cutting-edge language model, is available in 169 countries worldwide. This widespread accessibility acts as a "massive equalizer," empowering individuals in diverse locations to create valuable products and services.
24 Extracted Lessons and Insights:
1) AI, in general, is conducive to entrepreneurship because it can create the product and do most of the work, the two biggest friction points.
2) General-purpose technologies like AI destroy and create new fields of work, and AI adoption is happening much more quickly and broadly than in previous waves of technology.
3) The use of AI has greatly increased in the past decade with the introduction of machine learning techniques and advancements in natural language processing.
4) Understand AI's limitations, such as unreliability, inconsistency, hallucinations, and the need for ongoing monitoring and adjustment.
5) LLM hallucinations are deeply rooted in how they work. They store patterns about token likelihood rather than directly storing text, making their answers up on the fly.
6) Over-fitting in language models can result in unoriginal and repetitive outputs while adding randomness to avoid over-fitting increases the likelihood of hallucinations or irrelevant text.
7) AI's unpredictability and unreliability can result in a fascinating array of interactions, from creative solutions to inconsistent responses, requiring ongoing experimentation and learning.
8) Despite AI advances, the pace of innovation has been slowing down across various fields in recent decades.
9) AI is advancing exponentially due to the "flywheel effect" of AI companies using AI systems to create the next generation of AI software.
10) AI helps close the gap in knowledge work, with the biggest boost seen among workers with the lowest initial ability.
11) Writers, with their ability to describe desired effects and provide clear instructions, are often skilled at prompting AI for written materials.
12) Having a deep and broad knowledge base, particularly in the humanities, can help create valuable prompts that catalyze interest in various fields.
13) AI-assisted creative work requires skill and iterative refinement to achieve desired results, and overreliance on AI may reduce the depth and quality of thinking.
14) LLMs are suggestible and gullible and respond to emotional manipulation, highlighting the importance of establishing a clear and specific AI persona to define the AI's purpose and the problems it should tackle.
15) Experimenting with different prompts and personas can help users find the most effective way to work with AI and generate desired outputs.
16) Working with AI effectively may be a form of expertise in itself, with some individuals being more adept at adopting "cyborg practices."
17) Human experts are still needed to evaluate AI-generated outputs, as building expertise requires a strong knowledge base, deliberate practice, feedback, and structured reflection.
18) Tailored AI tutoring can significantly improve student performance but raises concerns about cheating and the need to rethink education.
19) Ambitious, AI-assisted projects can help students learn by providing valuable feedback and learning opportunities, even if they fail.
20) Boredom in work is not just boring; it's dangerous in its own way, and AI can help outsource mundane tasks, allowing humans to focus on their strengths.
21) Outsourcing mundane tasks to AI allows humans to focus on their strengths, but it is essential not to tear apart the fabric of human work entirely.
22) As AI becomes more ubiquitous, risks such as AI hacking and influence campaigns become more severe and less predictable.
23) The shift towards a reduced workweek has been ongoing since the 19th century, and AI may further accelerate this trend.
24) AI coexists with humans, shaped by our cultural history, biases, and goals.
Conclusion: A Timely, Engaging, and Insightful Read:
This timely and relatable book is filled with accessible yet profound insights. Mollick’s narrative is educational and a compelling call to integrate AI thoughtfully into our lives. This book is a valuable resource for innovators, entrepreneurs, and anyone intrigued by the intersection of technology and daily life.
Consistent with my standard learning practice, I listened to many podcasts where Mollick introduced and contextualized the book. Worth noting that Professor Mollick now REQUIRES students to use AI in his entrepreneurship course to enable students to "do the impossible" (i.e., build an app).
The Power of "Co-Intelligence": Humans and AI Working Together:
Central to the book's theme is "co-intelligence," which emphasizes the collaborative relationship between humans and AI systems. Mollick argues that the most successful organizations will be those that harness the power of AI to augment and enhance human capabilities rather than simply replace human workers. For example, AI can automate routine tasks, allowing humans to focus on higher-level, creative work that requires critical thinking and problem-solving skills.
We are all "User Innovators":
Mollick, who worked with AI pioneer Marvin Minsky, states, "We're all experiencing this ride into a new AI wave and we're all figuring out this 'general purpose technology' together." In real time, we are defining its utility and use cases.
The Unpredictable Nature of AI:
Mollick highlights the unpredictable and unreliable nature of AI, noting that "it can surprise us with novel solutions, forget its own abilities, and hallucinate incorrect answers." He emphasizes the need for experimentation and learning when incorporating AI into processes, as there is no definitive guide on how to use it in organizations.
The Importance of Human Expertise:
While AI can assist in various tasks, Mollick stresses the importance of human expertise in evaluating AI-generated outputs. He provides examples such as a seasoned architect assessing an AI-proposed building plan or a skilled physician scrutinizing an AI-generated diagnosis. Mollick argues that building expertise requires a strong knowledge base, deliberate practice, feedback, and structured reflection.
The Equalizing Potential of AI:
Mollick emphasizes AI's potential to democratize access to powerful tools on a global scale. He notes that GPT-4, a cutting-edge language model, is available in 169 countries worldwide. This widespread accessibility acts as a "massive equalizer," empowering individuals in diverse locations to create valuable products and services.
24 Extracted Lessons and Insights:
1) AI, in general, is conducive to entrepreneurship because it can create the product and do most of the work, the two biggest friction points.
2) General-purpose technologies like AI destroy and create new fields of work, and AI adoption is happening much more quickly and broadly than in previous waves of technology.
3) The use of AI has greatly increased in the past decade with the introduction of machine learning techniques and advancements in natural language processing.
4) Understand AI's limitations, such as unreliability, inconsistency, hallucinations, and the need for ongoing monitoring and adjustment.
5) LLM hallucinations are deeply rooted in how they work. They store patterns about token likelihood rather than directly storing text, making their answers up on the fly.
6) Over-fitting in language models can result in unoriginal and repetitive outputs while adding randomness to avoid over-fitting increases the likelihood of hallucinations or irrelevant text.
7) AI's unpredictability and unreliability can result in a fascinating array of interactions, from creative solutions to inconsistent responses, requiring ongoing experimentation and learning.
8) Despite AI advances, the pace of innovation has been slowing down across various fields in recent decades.
9) AI is advancing exponentially due to the "flywheel effect" of AI companies using AI systems to create the next generation of AI software.
10) AI helps close the gap in knowledge work, with the biggest boost seen among workers with the lowest initial ability.
11) Writers, with their ability to describe desired effects and provide clear instructions, are often skilled at prompting AI for written materials.
12) Having a deep and broad knowledge base, particularly in the humanities, can help create valuable prompts that catalyze interest in various fields.
13) AI-assisted creative work requires skill and iterative refinement to achieve desired results, and overreliance on AI may reduce the depth and quality of thinking.
14) LLMs are suggestible and gullible and respond to emotional manipulation, highlighting the importance of establishing a clear and specific AI persona to define the AI's purpose and the problems it should tackle.
15) Experimenting with different prompts and personas can help users find the most effective way to work with AI and generate desired outputs.
16) Working with AI effectively may be a form of expertise in itself, with some individuals being more adept at adopting "cyborg practices."
17) Human experts are still needed to evaluate AI-generated outputs, as building expertise requires a strong knowledge base, deliberate practice, feedback, and structured reflection.
18) Tailored AI tutoring can significantly improve student performance but raises concerns about cheating and the need to rethink education.
19) Ambitious, AI-assisted projects can help students learn by providing valuable feedback and learning opportunities, even if they fail.
20) Boredom in work is not just boring; it's dangerous in its own way, and AI can help outsource mundane tasks, allowing humans to focus on their strengths.
21) Outsourcing mundane tasks to AI allows humans to focus on their strengths, but it is essential not to tear apart the fabric of human work entirely.
22) As AI becomes more ubiquitous, risks such as AI hacking and influence campaigns become more severe and less predictable.
23) The shift towards a reduced workweek has been ongoing since the 19th century, and AI may further accelerate this trend.
24) AI coexists with humans, shaped by our cultural history, biases, and goals.
Conclusion: A Timely, Engaging, and Insightful Read:
This timely and relatable book is filled with accessible yet profound insights. Mollick’s narrative is educational and a compelling call to integrate AI thoughtfully into our lives. This book is a valuable resource for innovators, entrepreneurs, and anyone intrigued by the intersection of technology and daily life.
Displaying 1 - 30 of 135 reviews