Baley is called to the planet Aurora to investigate the "murder" of Jander Parnell, a humaniform robot — and one of only two in existence, the other being Baley's partner Daneel.
More so than The Caves of Steel (which I thoroughly enjoyed), this novel is an extended exercise in logic. Right near the start, in fact, there's a pretty intense discussion regarding whether a robot could be said to be "alive," and whether a robot could then be said to have been "murdered," a victim of roboticide. Generally, one would say a robot is destroyed, its usefulness terminated, and should it in questionable circumstances, the crime would be considered an act of vandalism. Here, the humaniformity of the robot is a factor in taking the case to another level.
The novel is almost all talk, dialogue, in the form of cross-examination and explication, and very little action. Asimov, through the Baley character channels Hercule Poirot, in conducting the investigation generally, and Raymond Smullyan, in negotiating the paths of logic related to the Three Laws of Robotics.
While Smullyan's logic puzzles may make my brain hurt, the type of paradox in which complex instructions given a robot are in contradiction of one or more of the Laws — and sorting out the the relative worth of defying an order versus breaking a Law — could immobilize a robot, or cause it to short-circuit.
Baley said, "And something like this, I take it, was what happened to Jander Parnell. He was faced with a contradiction in terms and his brain burned out?"
"It's what appears to have happened, though that is not as easy to bring about as it would been in Susan Calvin's day. Possibly because of the legend, roboticists have always been careful to make it as difficult as possible for contradictions to arise. As the theory of positronic brains has grown more subtle and as the practice of positronic brain design has grown more intricate, increasingly successful systems have been devised to have all situations that might arise resolve into nonequality, so that some action can always be taken that will be interpreted as obeying the First Law."
"Well, then, you can't burn out a robot's brain. Is that what you're saying? Because if you are, what happened to Jander?"
"It's not what I'm saying. The increasingly successful systems I speak of, are never completely successful. They cannot be. No matter how subtle and intricate a brain might be, there is always some way of setting up a contradiction. That is a fundamental truth of mathematics. It will remain forever impossible to produce a brain so subtle and intricate as to reduce the chance of contradiction to zero. Never quite to zero. However, the systems have been made so close to zero that to bring about a mental freeze-out by setting up a suitable contradiction would require a deep understanding of the particular positronic brain being dealt with — and that would take a clever theoretician."
It's a very entertaining novel and not at all what I expected. It reminds me a little of Stanisław Lem in its simple compelexity, following a set of basic premises to their logical conclusion.
Oh! And! There's robot sex! Or at least, human–robot sexual relations. On a planet where everyone sleeps with everybody else (including fathers and daughters). Which is, in the context of the conditions under which humans colonized the galaxy, all very plausible speculation as to how society and its morals might evolve in the circumstances of a given planet.