Final four – out of seven

 kw: story reviews, science fiction, fantasy, anthologies, collections

Three stories of the last seven in The Best American Science Fiction and Fantasy 2025, guest edited by Nnedi Okorafor, that I rated, "OK, there's an interesting idea or two here," and there is one I liked well enough to say, "I'm glad I read it." The other three were quickly rejected; stuff I don't wish to burden my memory with.

The "interesting" three are 

• "Reduce! Reuse! Recycle!" by T.J. Klune. At the end of ten years of uninterrupted labor in a factory, a semisentient robot is given one week to experience life outside of work. Then it is to be recycled by having its memories erased and getting equipment and software upgrades. Another ten year stint is to follow. Et cetera. I confess that I skimmed past much of the day by day details of the last week that "Douglas", or P-23, experienced. It is the somewhat predictable rollup of sightseeing, new experiences, making friends, and so forth. The crux is the final discussion Douglas has with the Supervisor. As robot technology improves, so does their self-awareness. Douglas does indeed get recycled, but I'll leave it to you to read for yourself. It points up the ethical issues surrounding AI, should it ever achieve consciousness, but I'll return to this theme later on.
• "The Sort" by Thomas Ha. In a world that embraced genetic modification far beyond what we have so far achieved, and then began to reject it, a man and his son are sightseeing, driving in the desert Southwest. They are far from their home, where their own modified humanness is known and probably accepted (this isn't quite clear), and now among people who don't all tolerate modified humans. The man and his son are telepathic, but only with each other. There are hints that their telepathy extends to certain other modified entities, such as a Tiger that can speak a few words, and perhaps even the basketball-sized garlic bulbs to be sorted in a strange ritual. I would ordinarily pass off such a story as "going nowhere," but in this case that is the point. Having pulled back from rampant modification-for-modifications'-sake, most of the world has thrown out the baby with the bathwater and is stuck in a reactionary rut. To do something "just because you can" is frequently the worst reason for doing it.
• "Ushers" by Joe Hill. Many cultures, including American folklore, contain a grim reaper or banshee or similar supernatural entities that either accomplish a person's death or foretell it or, in the case of this story, accompany the soon-to-die and "usher" them into the afterlife. In this story, a young man, Martin, can see the Ushers, which makes him act strangely at times, and his history of avoiding certain disasters has piqued the attention of the FBI. Most of the story consists of two interviews. Knowing beforehand that his interviewers are soon to be ushered off the scene, Martin tells of his "gift" to one of them. Soon after, they are Ushered out of this life in an accident.

The story I liked best is "The Three Thousand, Four Hundred Twenty-Third Law of Robotics" by Adam-Troy Castro. The title together with the content of the story gives a better view of the complexity of the ethical and moral intricacies that will be needed to produce wholly beneficial AGI (G for General) or ASI (S for Super), whether in a robot body or not. The story consists of one enormous sentence, extremely well constructed to avoid the turgidity of single-sentence legal contracts. A self-aware robot has been placed on a planet by its cruel owner, ordered to await his return without moving, and abandoned. It knows this. 

If for no other reason, it is worthwhile to get this volume for this story alone, and to read this incredible sentence thoughtfully, in detail. I will mention just one item, the core idea of human-AI ethics: If a mechanism is created that can feel pain, there are some (many?) people who will inflict pain on purpose. This is a thesis also found in the opening scenes of Caves of Steel by Isaac Asimov. Furthermore, insomuch as a mechanism can be used to cause pain in other persons, there are some (many?) who will do so, apparently with great pleasure to themselves.

This is part of the reason I refer to AI as SI, Simulated Intelligence. Part of the reason I dearly hope that sentience is forever impossible to mechanisms.