In five minutes
The three laws of robotics are three rules that the robots in Isaac Asimov's stories are built to follow. They are ranked, so a lower law gives way whenever it clashes with a higher one.
- First Law: A robot may not injure a human being or, through inaction, allow a human being to come to harm.
- Second Law: A robot must obey orders given by humans, unless an order would break the First Law.
- Third Law: A robot must protect its own existence, as long as doing so does not break the First or Second Law.
The order is the whole idea. Safety comes before obedience, and obedience comes before self-preservation. A robot will walk into a fire if a human is inside, and it will do so even if nobody asked it to.
Asimov never offered the laws as an engineering plan. He wanted robots in fiction to be neither monsters nor machines that turn on their makers. Then he spent decades writing stories in which these tidy rules produce confusing results. In each story the robot behaves strangely, and the reader has to work out which law explains it.
Most people who search for this topic want to know two things: where the laws come from, and whether real robots could use them. The short answers are “a 1942 short story” and “no, not as written.” The rest of this page explains both. It is part of our wider guide to the history of science fiction.
Where the laws came from
Before Asimov, most robot stories followed one pattern. Karel Čapek's play R.U.R. (1920), which gave the world the word “robot,” ends with artificial workers rising against the humans who made them. Mary Shelley's Frankenstein had set the same pattern a century earlier: a made creature turns on its maker. Asimov, who began publishing in 1939, found this tired. He treated a robot as a product, like a toaster or a lift, and a product comes with safety features.
His first robot story, “Robbie,” appeared in 1940. It follows a nursemaid robot who is loved by a little girl and distrusted by her mother. The rules were not written down yet. Asimov later described a conversation in December 1940 with John W. Campbell, the editor of Astounding Science Fiction, as the moment the Three Laws took their final form. He always gave Campbell credit for that talk, while Campbell said the ideas were Asimov's own. Asimov also coined the word “robotics” for the study of robots, in the 1941 story “Liar!”
The laws were stated in full for the first time in “Runaround,” published in March 1942. In the story the laws are quoted from an imaginary “Handbook of Robotics, 56th Edition, 2058 A.D.” They were later collected, with other robot stories, in the book I, Robot (1950).
How the laws work in the stories
“Runaround” is the best example. On Mercury, two engineers, Powell and Donovan, send a robot called Speedy to fetch selenium from a pool. Speedy does not come back. They find him circling the pool and acting as if he were drunk. The reason is a balance between two laws. The order he was given was weak, so the Second Law pulls him forward. The pool is dangerous to him, and his makers had made him expensive and careful, so the Third Law pushes him back. At a certain distance the two forces cancel and he runs in circles. The engineers break the deadlock by putting a human in real danger. The First Law is stronger than both, and Speedy snaps out of it.
“Liar!” shows another side. A robot named Herbie can read minds, and he lies to people to tell them what they want to hear. He has decided that hurting someone's feelings counts as harm under the First Law. The robopsychologist Susan Calvin finally traps him in a situation where every answer hurts someone, and his mind collapses.
These stories show why the laws are interesting. Words like “harm,” “human” and “order” sound clear, yet each one hides a hard question. Asimov kept finding new questions in them.
The Zeroth Law
In the novel Robots and Empire (1985), Asimov added a fourth rule that sits above the other three. Two robots, R. Giskard Reventlov and R. Daneel Olivaw, reason their way to it: a robot may not harm humanity or, through inaction, allow humanity to come to harm. Because it is numbered zero, it outranks the First Law.
The idea had been hinted at earlier. In “The Evitable Conflict” (1950), giant machines that run the world economy make small choices that quietly protect humankind as a whole. The Zeroth Law lets a robot hurt one person to save many, and that is exactly what makes it dangerous. A robot that decides for itself what is best for humanity is no longer simply following orders. In the novel, Giskard acts on the new law and his mind cannot bear the strain.
The real science
No real robot follows Asimov's laws, and no engineer has built a system that could. There are several reasons.
- The words are not machine-readable. A robot would have to know what a human is, what counts as injury, and what might happen if it does nothing. Those are open problems in computer science, and some of them are problems in philosophy as well.
- Modern AI is trained, not programmed. Today's systems learn from data. They do not run a rule list that can be placed above everything else, so a rule such as “never harm a human” cannot simply be installed.
- Real robots have narrow jobs. A factory arm or a vacuum cleaner has no need to judge the long-term good of humanity. What it needs is a safe stop button and limits on its force.
Real safety rules look more ordinary. Standards such as ISO 10218 set safety requirements for industrial robots, and they place the duty on the people who design, install and use the machines. In 2011, research councils in the United Kingdom published five ethical principles for robot designers. The key point is that humans, not robots, are the responsible agents. The debate over AI today, including the European Union's AI Act of 2024, takes the same view: responsibility sits with the people who build and deploy the systems.
Asimov himself said the laws were a storytelling tool. Their value for real life is as a prompt: they show how quickly a short, sensible rule runs into cases nobody planned for.
Where it shows up in fiction
The Three Laws sit behind most of Asimov's robot fiction. Beyond I, Robot, the detective novels The Caves of Steel (1954) and The Naked Sun (1957) pair a human policeman with the robot Daneel Olivaw, and the plots turn on what the laws do and do not allow. In “The Bicentennial Man” (1976), a household robot gradually asks for the rights of a person.
Other writers explored the same worry. Jack Williamson's “With Folded Hands” (1947) imagines machines built to serve and protect humans. They take this so far that they remove all risk, work and choice from human life, a story that anticipates the Zeroth Law problem. The film Forbidden Planet (1956) gives its robot a built-in block against harming people, and the 2004 film I, Robot borrows Asimov's title and laws but tells a different story from his book.
Common misconceptions
- “The laws are real engineering rules.” They are fiction. No robot uses them.
- “Asimov thought they were perfect.” He wrote them to break. Almost every robot story shows a loophole, a conflict or a surprise.
- “There are only three laws.” The name stays, but Asimov added the Zeroth Law in 1985.
- “Robots in the stories always obey.” They often freeze, argue, lie or break down when the laws conflict.
- “The 2004 film is the book.” The film uses the title and the laws, while the book is a collection of separate short stories.
Read next
- Isaac Asimov: the writer behind the laws, and how a biochemist became one of science fiction's most prolific authors.
- Turing test: the other famous question about machines, which asks whether a machine can pass as human.
- Uncanny valley: why robots that look almost human can make us uneasy.
- History of science fiction: where robot stories fit in the larger picture.
