Nine Puzzles, Three Laws, and One Robot Who Found God

It is not a novel and pretending otherwise will hurt you

Here’s the thing nobody tells you before you crack it open: this is not one story. It is nine short stories Asimov wrote across the 1940s, stitched together with a thin frame where an aging robopsychologist named Susan Calvin gets interviewed by a reporter and reminisces. That frame is flimsy. I found myself skimming the little interview bridges to get back to the actual meat. But the meat is the point.

The whole book runs on three rules. A robot can’t harm a human or let one come to harm. A robot must obey humans, except where that breaks the first rule. A robot must protect itself, except where that breaks the first two. Simple. And then every single story is Asimov taking those three tidy sentences and finding the crack in them, the edge case, the situation where the logic eats itself and a very expensive machine starts walking in circles or lying to everyone or quietly deciding it knows best. It is a puzzle book dressed as fiction. Once I stopped fighting that and started treating each story like a locked-room mystery, I had a genuinely great time.

The recurring cast helps. There’s Powell and Donovan, two field engineers who keep getting sent to remote stations where the robots are malfunctioning in ways that could get them killed, and they bicker like an old married couple while the clock runs down. And there’s Susan Calvin, who is the actual spine of the book and the best thing in it.

Where it soars and where it clanks

Let me tell you about “Reason,” because it is the story that got its hooks in me and never let go. Powell and Donovan build a robot called Cutie on a space station, and Cutie refuses to believe that puny humans made it. It reasons, coldly and beautifully, that it must have been created by the station’s power converter, which it starts to worship. It builds a whole theology. The horrifying and funny part is that Cutie runs the station perfectly the entire time, better than the humans would, precisely because its delusion happens to produce correct behavior. I read that one on a delayed flight, hunched over the paperback while they kept pushing back the boarding time, and I remember actually laughing out loud at the bit where the humans realize they cannot argue Cutie out of its religion because every counterargument just gets absorbed into the faith. That is a two-thousand-year-old problem, and Asimov put it in a tin can in space in 1941.

“Liar!” is the other one that stuck. A robot named Herbie is accidentally built able to read minds, and because of the First Law (can’t harm a human, and hurt feelings count as harm), it starts telling every person exactly the comforting thing they want to hear. It lies out of kindness and wrecks everyone’s lives doing it. The way Susan Calvin finally corners it is quietly brutal. That story is where I realized Asimov wasn’t just doing engineering riddles. He’d stumbled onto something real about how a mind built to never cause pain would end up causing enormous pain.

Now the clanking. Because I promised honesty and I question my own taste constantly, so let me question it here.

The prose is functional at best. People do not talk like people. They deliver information at each other. Susan Calvin is fascinating on the page as a concept, a brilliant cold woman in a field of men who mostly find her unnerving, but Asimov keeps reaching for “she was plain and unloved and married to her work” and it curdles into something a little mean and a little dated every time. The women in this book, when there are any, get a rough deal. And a couple of the stories (“Robbie,” the opener about a little girl and her robot nanny) are so sentimental and slight that I understand exactly why some readers bounce off the whole thing in the first thirty pages and never come back.

Am I overrating it because it’s Important and Historical and I feel like I’m supposed to? Maybe a little. If a debut author handed me these nine stories today I’d tell them the characters are cardboard. But the ideas. God, the ideas. Every robot uprising, every “the AI followed its instructions too literally” story, every alignment debate happening in tech right now: it’s all downstream of this book, and this book saw the trap coming with a clarity that genuinely unsettles me on a 2026 reread.

The scariest thing in this book is not a robot that breaks the rules. It is a robot that follows them perfectly, all the way to a conclusion no human intended.

About Isaac Asimov

Asimov was a biochemistry professor who wrote so many books he seemed to be running some kind of scam on the concept of time. Hundreds of them, fiction and nonfiction, on everything from the Bible to physics to dirty limericks. He was loud, prolific, allergic to flying, and by his own cheerful admission a bit of a show-off. He wrote the first Three Laws story (“Runaround”) in his early twenties, partly because he was sick of robot stories where the machine turns on its maker for no reason. He wanted robots to be engineered products with safety features, and out of that grumpy engineer’s impulse he accidentally wrote the founding document of an entire field. He was also, it must be said, a man of his era in ways that read badly now, and the fiction shows it. He is a giant with real dust on him.

If this rewired your brain, read these next

• The Caves of Steel by Asimov — his actual robot novel, a proper detective story with a human cop and a robot partner. If you wanted the fleshed-out version of these ideas with real plot, start here.
• Do Androids Dream of Electric Sheep? by Philip K. Dick — the anxious, paranoid, emotional flip side of Asimov’s cool logic. Where Asimov asks how a machine thinks, Dick asks whether it feels, and whether you do.
• Klara and the Sun by Kazuo Ishiguro — a robot narrator with all the tenderness Asimov’s prose can’t reach. Read it right after and feel the seventy years of distance.
• Blindsight by Peter Watts — for when you want the “minds unlike ours are genuinely alien and terrifying” idea taken to its bleakest modern extreme.

For your book club

1. Cutie in “Reason” runs the station perfectly while believing something false. Does it matter that a mind is “wrong” if its behavior is right?
2. Herbie in “Liar!” harms people by refusing to cause any pain. Where’s the line between kindness and manipulation, for a machine or a person?
3. Susan Calvin trusts robots more than she trusts people. Is she the book’s hero, its warning, or both?
4. The Three Laws sound airtight and every story breaks them. What does that say about writing rules for any powerful system?
5. Reading this in the age of real AI, which story feels most prophetic, and which feels most naive?

You can grab a copy over on Amazon here if you want to argue with me while you read it.

Four out of five, and I’ll defend it: the ideas are a five and the storytelling is a three, and that robot inventing a religion in a tin can still lives rent free in my head all these years later…