Blindsight
Blindsight
Copyright © 2006 by Peter Watts
A Tor Essentials book
ISBN 978-1-250-23748-4
I decided to read Blindsight based on this blog post, titled Blindsight and Gorgias: The Chinese Room and Sophistry.
I wish I could say that the reviewer hooked me with the explanation of the Chinese Room thought experiment, or comparisons to Neal Stephenson’s books, which I, too, have read and enjoyed. It was this sentence that compelled me to read Blindsight:
It reminds me of how some isolated internet autistics describes their experience in trying to communicate with neurotypicals in social situations, an impossibly large set of rules to memorize and apply that they were never privy to.
That was pretty much how I’ve felt my entire adult life. I phrased it a little differently, I didn’t quite get the concept of a set of complicated rules. I used to say that I’d somehow missed the Secret Handshake that someone taught most people, and whatever magic the Secret Handshake conveyed allowed most people to interact effortlessly in social situations.
That said, I did not like Siri Keeton the protagonist. Even I know he did his ex, Chelsea, incredibly wrong. Depicting Siri as able to detect subterfuges and subtle emotional tics, but not giving him the ability to do anything with that knowledge was infuriating.
This book is great science fiction. Humans have become a Kardashev Type I civilization, expanding to Earth’s Moon and Mars. They’re capable of resurrecting a human subspecies, called “vampires” in the book, and sending a crewed probe into the Oort Cloud. The author Peter Watts basically sheds good ideas throughout it. The story is not at all optimistic, not like Double Star or Mission to the Moon. It’s a product of the early 2000s, just like those books are products of the 1950s.
The alien entity, Rorschach, is possibly more believable today that in 2006 at the time Blindsight was written. The few conversations the humans have with Rorschach are definitely LLM-scented. The AGI that runs the ship Theseus is perhaps less believable today than in 2006. It seems entirely possible that “AI” will never be AGI, and after getting prompted, will occasionally produce absolute rubbish.
Possible Spoilers
If The Captain, who was Theseus, a “quantum computer” AI, ran the entity named Jukka Sarasti, why did The Captain torture Siri into regaining some empathy, into losing his objectivity?
Sarasti was portrayed as inscrutable and aloof. As a vampire, a human predator, he could easily have engaged in some animal husbandry. It doesn’t entirely make sense for an AI, a real general AI, not some pattern matching, statistical LLM, to try to jolt Siri into feeling things.
The names of non-human entities are also rather on-the-nose after you’ve finished the book “Theseus” seems like a great thing to name a ship that can produce anything you want if you’ve got the quantum (or otherwise) specs for your item.
“Rorschach” brings to mind both the Watchmen character who was so nuts he many not have been conscious, and the famed inkblots which deliberately have no intrinsic meaning.
Finally, the scramblers. The human linguist(s), “The Gang of Four”, try to get known-text cipher text out of two scramblers by hurting one. The scrambler being tortured has to communicate to the other scrambler a shape or color or something, which the second scrambler can indicate to stop the torture.
If the scramblers are not conscious, why would the first scrambler try to avoid the pain? Why would the second scrambler bother to try to relieve the pain felt by the first scrambler? I found that part a bit gimmicky. It did not ruin the plot, the book, or the philosphizing.