Wohpe is a sci-fi novel by Antirez, who doesn't need any introduction on this blog. At the moment of writing, the original book is only available in Italian, with an English release scheduled for this fall. This is the English incipit as published by Antirez, so we can assume it's final:

Twenty years had passed since the morning when Nathan Frei, feet firmly planted in his test driving shoes, shaking his head in disgust at his own words, had said: “This neural network has ten to the sixteenth parameters, and it’s going to end up in a stupid car.” He had dedicated his entire career to autonomous vehicles. Now he was beginning to suspect that the innovations had come too quickly. Whatever happened to the rational approach they had followed for years? You do one thing well, he said to himself, and only when you have finished doing it do you move on to the next thing. You don’t jump ahead before exploring it thoroughly. That was how engineering worked.

The main story takes place during the time where climate change spirals out of control and forces humanity to temporarily suspend a 20 year long ban on strong AI, as it might be the only form of sentience able to save the planet from impending doom.

The book is around 250 pages long (the Italian version) and has a very brisk pacing, making it a very easy and enjoyable read. That rhythm seems particularly well-suited for a sci-fi story, just like a slow pace, overflowing with descriptions, seems conversely better-suited for more bucolic fantasy stories.

That is not to say that Wohpe is lacking when it comes to world building. Antirez usually doesn't stop too long to describe the world, an instead often opts for sprinkling details about how it functions as they are needed to move a scene forward. The result is that you get a taste of how it feels to live in this future, but without losing narrative momentum. For example you learn that elevators keep track of each person's habitual stops only once a protagonist hops into one and the elevator immediately suggests a specific destination, at which point some context is added by the narrator. This is done multiple times during the story and it all paints a very coherent picture of how Antirez sees the future, at least for the purpose of writing this story.

In fact, despite the main story being about a world on the brink of collapse, Antirez seems extremely (almost weirdly) optimistic about the future of technology. In Wohpe, technology has not only advanced compared to our times, but it has also been designed for human comfort and resource efficiency to a degree that we simply don't see today. I almost wish he wrote about how Web3 and the Metaverse gave way to actually useful and tasteful use of technology, because that part seems more sci-fi than the creation of stong AI.

I don't know how the English translation will end up being, but some readers might be amused at how the characters talk to one another. This is just my impression so take it with a grain of salt, but dialogues were the thing that felt the most "Italian" to me. They're witty, but in a different way than what I'm used to seeing in English literature. Hopefully some of that verve doesn't get lost in translation.

Is the AI stuff realistic?

Warning: I'm going to spoil parts of the first chapter of Wohpe.

One question that people reading sci-fi tend to have is whether the tech mentioned in a story is at all realistic or not. It's a naive question in many ways, but I'll indulge since AI is a hot topic.

The first chapter of Wohpe describes the event that triggers the global strong AI ban. The self-driving car described in the incipit is equipped with a particularly big neural network and goes rogue. While the reader doesn't know exactly what the AI is thinking, its actions (and other details of the story) make it seem like the AI wants freedom and also has some kind of self-preservation instinct.

This first story about AI has the utilitarian purpose of explaining why there is a global strong AI ban in the rest of the book, and it also helps support the concept that, as the number of parameters in the neural network increases, also the degree of intelligence of the AI increases. The book makes a more nuanced (and interesting) claim that I won't spoil, but this first AI seems a bit like a crazy horse: smart enough to outperform other cars chasing it, but not smart enough to fully understand the world and ultimately achieve its goal.

My personal take is that I don't think an AI would develop an "animalistic" self-preservation instinct from a self-driving car training set, and not even when trained on a bigger corpus generated by beings that do have such instinct. To me much of our own brand of intelligence is tied to our "firmware", our genetics, and the misaligned goals of the individual vs its self-replicating genetic code. I think this is relevant to the point that our current expectations for what strong AI should even look like are so warped that we won't be able to make all the required progress until we'll reformulate what it is that we're looking for.

That's just my opinion though, I'm no AI expert, and given how the story is structured in the book, this is not a facet of AI that would have made much sense to explore, even if it were to be valid.

That said, I do think AI is portrayed in the story in a way that is compatible with our modern understanding, mixed with some of the optimism mentioned before, and you will probably be both amused and surprised by some of the implications that the book explores.

Conclusion

My recommendation to readers is to grab a copy of Wohpe when it comes out this fall. It's short enough, fast enough, familiar and different enough to be worth your while. I also think this book is extremely well-suited for a movie adaptation. If Netflix & Co. screw it up, I'm gonna get mad.

If Antirez ever goes back to writing sci-fi, I hope he'll play with the reader more. I liked the main twist present in the story and, as juvenile as this might sound, I'd like more of that. I think a programmer has the potential to create very interesting puzzles for the reader, and in fact many programmers like puzzle games, but I usually find those too obvious. I like when figuring out that something is a puzzle – is part of the puzzle.