"Co-Intelligence" by Ethan Mollick & "I, Robot" by Isaac Asimov
One book published in 2024 about living and working with AI, another book published in 1950 about living and working with robots.
It was coincidence that I was reading these books simultaneously. I certainly hadn’t planned to bundle them in one review. But the more I read, the uncannier the connections I was making. It’s also fascinating to me that both authors have “rules” for their topics:
Mollick’s Four Essential Rules for Integrating AI into Work and Life
Always invite AI to the table
Be the human in the loop
Treat AI like a human (but tell it what kind of human to be)
Assume this is the worst AI you will ever use
Asimov’s Three Laws of Robotics
A robot may not injure a human being, or, through inaction, allow a human being to come to harm.
A robot must obey the orders given it by human begins except where such orders would conflict with the First Law.
A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.
Co-Intelligence
Mollick’s book is to the Generative AI era what Nicholas Negroponte’s Being Digital was in the mid-1990s. (If you never read Negroponte’s book you were either too young or now it’s too late. But it was formative to my understanding the digital world in the early 1990s.) Mollick’s book not only explains AI to the layman (along with concepts like large language models, hallucinations, and the Jagged Frontier), he does it in a way that’s approachable. Non-MIT grads should have no problem reading and understanding him.
There’s a lot of talk about AI in my industry. And after reading Mollick’s book, I feel like most of what I’m hearing is just that – talk. Conjectures, worries, worst-case scenarios. None of it is very useful. But Mollick not only makes sense of AI, he shows ways to use it that should make armchair conspiracy theorists and luddites alike think, “Well, that makes sense.” The low-level example I frequently use is that I uploaded my AT&T bill to ChatGPT and told it to find ways of saving me money, and in less than a second I had nine solid suggestions. Again, this is pretty low-level, but one I don’t think most people are taking advantage of. The “Jagged Frontier” is the idea that we don’t know what AI is good at until we use it. (Don’t like AI images? Then don’t use it to create your images. Or maybe refine how you’re using it.)
I first discovered Ethan Mollick when MasterClass offered a three-part series on AI. If you have a subscription, watch it. If you don’t, it’s a great reason to subscribe. Just as James Clear’s MasterClass was an easy summary of his book Atomic Habits, it feels like everything in Co-Intelligence is captured in Mollick’s class.
I, Robot
Asimov is such a luminary in the sci-fi genre, even people like me who haven’t read about him know about him. I began this book expecting a novel, but found it’s actually a collection of short stories, very much in the vein of Ray Bradbury.
There are stories about robots serving as nursemaids, robots helping miners on other planets, and robots developing god complexes and refusing to believe they were created by humans. Asimov uses a framing device for his story, much like Bradbury does in The Illustrated Man. In Bradbury’s book, a traveler encounters a heavily tattooed man on the road and camps with him for the evening. As the Illustrated Man sleeps, the traveler watches his tattoos swirl into stories, which is how the otherwise disjointed collection of stories is anchored. In I, Robot, Asimov uses the character of Dr. Susan Calvin, a robopsychologist telling stories to a reporter.
But the real joy in I, Robot is seeing how Asimov uses each story to test his Three Laws of Robotics. The characters make frequent reference to them which grounds the reader and points out potential conflicts and contraries. In some stories they’re almost a trail of breadcrumbs the characters have to follow to understand why their robots are acting they way they are.
It’s interesting and even a little cute to read a sci-fi novel when the years described are in our past, and we’re still nowhere close to the future it predicted. In Asimov’s future, interstellar and interplanetary travel is commonplace by the 2020s. And in one story the reader is meant to be astonished that it takes a hyper-intelligent robot only 30 minutes to read three books.
Synthesis
Asimov describes a lot of facets of AI but he’s doing it in the shell of metal and moving parts and smelling of oil. Mollick’s book describes the next frontier in computer programming that many of us are dealing with (and even worrying about) today. But there are multiple areas of overlap. For the sake of brevity, here’s a salient one:
Built into ChatGPT’s programming hierarchy is the mandate to please the user, which is why it can sometimes give false information if it thinks the user will be happy with the answer. One of the stories in I, Robot is about a robot that can read brainwaves and will lie to humans so it doesn’t hurt their feelings (breaking the First Law to not injure humans). In both cases, the underlying question is if it’s better to be lied to than hurt. And an excellent answer is in another sci-fi story. In Interstellar, Matthew McConaughey’s character asks the robot TARS why its honesty settings are set at 90% instead of 100%. TARS replies, “Absolute honesty isn’t always the most diplomatic, nor the safest form of communication with emotional beings.”
READ ASIMOV’S BOOK IF: You like sci-fi short stories. There’s a lot more to these stories than robots and rockets.
READ MOLLICK’S BOOK IF: You want to speak more intelligently about what AI is and what it can do. And if you want to see how you can use it to your advantage.
Trailer for the MasterClass hosted by Ethan Mollick.
You can also subscribe to Ethan Mollick’s Substack here.
BONUS CONTENT: Which AI is best at creating an image of Mollick talking to Asimov?
I thought it would be fun to include an AI-generated image of Mollick speaking with Asimov. As with most AI experiments, it’s didn’t go smoothly. I tried four different image generators. For reference, here are actual photographs of Mollick and Asimov:
Microsoft Copilot
Copilot wouldn’t create an image of actual people because it violated its content policy. So I used the following prompt to describe them: “Create a photo realistic image of a Wharton professor speaking with a science fiction writer. The Wharton professor should have a beard, dark hair, slightly balding, look in his 50s and olive complexion. The science fiction writer should be caucasian, have white unkempt hair with large mutton chop sideburns, and wear glasses.”
X’s Grok
Grok had no compunctions about generating an image of celebrities, but it couldn’t generate an accurate Mollick. In fact, its first few attempts generated several images of one Asimov talking to another Asimov. Even after I specified, “Create an image of Wharton professor Ethan Mollick,” this was the best it could do:
OpenAI’s DALL-E
ChatGPT gave me the following on the first try. I’m getting strong Harry Potter vibes from this.
OpenArt
Again, my first results were images of Asimov speaking to himself. I highlighted the head of one of the Asimovs and gave a second prompt describing Mollick: “a beard, dark hair, slightly balding, look in his 50s and olive complexion.” Apparently, I described Salman Rushdie.
