Archive of Our Own

Chapter Text

A few things to know:

1. These snippets are not original. The majority of them are based on the robot-AU stories in a Chinese stickman manhua series called 小綠和小藍 (Beryl and Sapphire) by Ocarina; the original comic in Chinese is freely available on 騰訊動漫 (Tencent Animation and Comics) here. When I say "based on," I mean that the conversation topics, themes, and ideas exchanged between the characters in those particular snippets are almost taken exactly as they are in the comic. The core ideas and their progression in those particular discussions between characters are not my own. I merely filled out what the stickman format of the comic left unspecified with my own imaginings, substituting in Sherlock and John: their actions as they say certain things, how they might say those certain things (phrasing, tone, etc.), background details to build up a more plausible, more complete world in which such events and conversations might arise, etc. And, if I'm being generous to myself, I would perhaps also claim credit as a translator, for the original comic is Chinese, after all, and I did draw upon my own literary sensibilities to bring them to life in English—in particular, the way I envisioned conversations between a "Sherlock" and a "John" that members of the fandom would hopefully not find too OOC.
2. That said, there are certain details, often to do with linguistic features of robot-Sherlock's speech, and also a few—only a few—snippet premises that I did come up with myself, which I thought made sense for Sherlock and John, specifically.
3. However, these snippets were something I produced from a long time ago (judging from my file history, it's been at least six years), and I don't remember exactly which snippets were completely unoriginal, and which were less unoriginal.
4. As mentioned in the previous point, and harking back to the title I've given this work, this was written at least six years ago, and, by logic, any chapters of the Beryl and Sapphire manhua that inspired—I know that's a generous way to phrase it—the snippets in this work would have been in existence before this work was even conceived. While I was reading through these snippets after I dug them out from my Google Drive, I constantly found myself sighing through my nose.
5. As someone who currently works in tech and (I'd like to think) understand, on more technical grounds, the general workings and limitations of AI right now—most relevantly LLMs—I have to say this: The way Beryl and Sapphire, and by extension this unoriginal fic of mine, portrays AGI (artificial general intelligence; think of it as an intelligent robot/agent capable of interacting with the environment around them, interfacing with us through human language) is not at all what we see happening in AI today, especially on the language front.
6. LLMs are nothing if not adept at mimicking human speech; that's basically what they're trained to do. They do not have an objective understanding of "truth." When you ask ChatGPT or Gemini a question, sometimes you get "hallucinations"—that's because these models are not trained to generate "correct answers," they're trained to generate what humans would think sound like "correct answers."
7. Instead of pursuing the development of intelligent machines that could serve as paragons of knowledge and truth and rationality, as Beryl and Sapphire and this fic might have presumed would happen many years ago, big tech has instead collectively been chasing after... well, whatever it is we now have with all these chatbots and AI image and video generators that try to approximate human behavior, language, and creations and whose goal is ultimately to "pass" as a human, fallacies and all, and not to be "right." That's why these chatbots hallucinate: why they're generally very agreeable, even when you insist on something objectively incorrect (sometimes they really don't know you're incorrect; you can use AI to get approximate answers and summarize information that's out there, but definitely don't use AI responses as your ultimate "sources"); why they can be such devastating vehicles of misinformation and disinformation; why they facilitate confirmation biases; why they cite papers and theorems and concepts and books that don't exist; why they might be dangerous to people struggling mentally; et cetera, et cetera. Should it really surprise anyone that it's always ultimately boiled down to profit, when it comes to big tech? Most people, I think, would much prefer an "intelligent" robot that agrees with whatever you say and doesn't call you dumb, and not one that acts like the Sherlock in this fic does. That is—and I say this with fondness and self-awareness—like some very smart high-functioning individual on the autism spectrum who often has trouble understanding emotions, finds social norms illogical and performative, and over-rationalizes (or so it seems to others).
8. So why did I decide to post this fic instead of letting it sink into the depths of oblivion in my Google Drive, soon to be forgotten once more? I guess it's because I thought it interesting that, just single-digit years ago, naive little high-school me thought such a portrayal of artificial intelligence was plausible. Look at where we are now. This really is an alternate-universe fic, isn't it? Alternate with respect to the Sherlock canon (whichever one), and, also, alternate with respect to the [INSERT ADJECTIVE HERE] world we find ourselves in right now.

And with all that, I guess, let's get into it.