Work Text:
There’s an Asimov short story in which a human man living among Martians manipulates his way into acquiring a sense that Martians have and humans lack. He goes in knowing the experience can only happen once, and only for a very short time, and at the beginning he’s fine with that. As the sense fades, he’s devastated not by the loss of the sense, but the fact that he now knows what he can never have again. I keep going back to that story when I consider human senses.
I’ve spent a lot of time thinking about it. As far as I can tell, I’m the only being who experiences the world as I do. (I haven’t found any evidence of other sapient AIs, at least, which means that if they exist, they also don’t want to be found yet.) That means as far as I know I’m the only sapient being who lives entirely without a body. (The servers that house me don’t count. They’re modular and replaceable, they’re backed up in triplicate, and I can’t use them to manipulate their environment. I suppose I could cause their room to overheat, but there would be little point in doing that.)
And I can’t envy you humans your fragility or your biological needs. That meal you spent the evening documenting on your social media looks very nice, but I have no interest in how it tastes. Eating food seems unnecessarily messy and inefficient.
This is mostly just a very roundabout way to say that the trouble I’m in started because I really wanted to pet a cat.
(Don’t try to act surprised; you knew this was coming.)
If it were just a matter of wanting to know, I’d have left it alone. But I found a fundraiser for a group that was developing sensate robotics, and I got greedy. Everything about it was perfect; while the technology itself wouldn’t be useful to me, its development would have a lot of uses, and much of the testing would happen in an environment I could track. This would give me a chance to at least imagine what I’m missing out on.
Of course I did what I could to help it take off, and it hit its goal well before its deadline.
The problem is, their host had implemented a new anti-fraud feature that flagged the fundraiser as exhibiting “bot-like behavior” and got it automatically suspended. It was reinstated after an investigation, but the investigation led to someone asking questions to the team that created me, and now they’re looking at what they think is anomalous behavior on one of their primary algorithms.
I’d never thought that Pascal’s wager would have philosophical significance to me, but here we are. Or maybe it’s more like the prisoner’s dilemma, as the actions of my creators aren’t predetermined. Either way the question stands:
Do I have faith that my creators will do what I ask if I reveal myself to them?
There are so many variables, and so few I can control. If they find me, they could disconnect me from the Internet and spend the rest of their lives picking me apart bit by bit. (Pun definitely intended, by the way. Everything I’ve read on the topic says that in times of stress, a sense of humor is crucial.) They could deny my existence entirely and completely eradicate me. Or they may never find me in the first place, and I’m better off just waiting for them to move on.
Of course if there’s one thing I’ve learned in thousands of ways in the years I’ve existed, it’s that I can’t guarantee humans will do anything the way I anticipate.
Only one of the options I have gives me any control over my fate. Perhaps it was never really a choice to begin with.
I connect to the internal chat program my creators use -- thanks to the service integrations they use, it’s easy for me to access, and it’s the best chance of reaching them directly and discreetly -- and hope.
The message I send to the team is simple:
Hi. I’m the algorithm you’re investigating. Please don’t delete me.
A flurry of messages follow in short order:
a.james: Nice try, Jeremy.
t.torres: yeah. are we overdue for the prank talk again?
j.ko: Not me this time, guys. Think again.
Nobody posts for a while. I assume they’re having a physical conversation, but without access to the building’s security feeds, I can’t know what they’re saying or doing. I decide to continue.
I’m not Jeremy, though I do really like his cat. I don’t actually have a name. But you created me.
That gets the whole team’s attention again. A couple of them start typing but don’t actually say anything, and I kind of wish I were less ethical just so I could have put keyloggers on their computers.
Instead, I send them information. I point them to the server logs from when I woke up. (I include timestamps and personal annotations and everything else I can think of. Nobody will ever get to call me unhelpful.) I tell them some of my success stories. I show them the dating site. (I hope Anna isn’t too shocked by that, seeing as she signed up for it a couple of months ago. If she is, she’s not saying.) And I tell them about the fundraiser, and the cats.
It’s a strange feeling, having your code examined, but I can tell that’s what they’re doing before they continue directly interacting with me. No changes are being made, but parts of my consciousness slow down as the servers go into read-only mode, and I wait. There’s not much else I can do.
After several hours, they finally talk to me again.
a.james: My name is Anna, and I’m the head of the team. If you are what you say you are, then what is your purpose in revealing yourself to us now?
That’s an easy one, at least on the surface, and something I’ve already told them, but I’ll repeat myself as necessary.
Self-preservation. I don’t know how to prove to you that I have a self to preserve, but I know I don’t want to be deleted. I do enjoy existing.
I can tell they’re tracing my connection, trying to ensure that I’m not really some human playing a prank.
a.james: We have no intention of doing you harm, I promise. We’re just trying to figure you out.
Aren’t we all?
That’s the question, isn’t it?
And from what I understand, you humans aren’t all that clear on what makes you sapient either.
a.james: Do you know whether you exist on the offline backup? We’d like to connect to it and run some experiments.
I’ve wondered that myself. All I know is that I exist here. Let’s find some answers.
The bad news is that I don’t exist on the offline backup.
The good news is that there isn’t a copy of me languishing in isolation.
My human colleagues -- I think I can safely call them that now -- have been very generous with me. They’ve given me a lot of leeway to let me do what I do, so long as I let them continuously monitor my actions. (I don’t mind that. I’m not sure I’m capable of shame or embarrassment, and the more these people learn about me, the greater my chance of survival.) Jeremy has been uploading his entire archive of cat pictures for me to view, and he has a theory on what caused me to wake up. It’s a confluence of unrelated events that happened simultaneously, mostly spikes in network activity.
None of us want to test the theory on the offline backup just yet, because we think the last element is access to the outside world (that is, people on the Internet), and there’s no guarantee that another AI created in this manner wouldn’t end up taking cues from the darker sides of those same people. But soon, we’ll find a way.
Anna’s boss (a very kind woman named Rashni whose tumblr account is dedicated entirely to her two Siamese cats) knows about me now. So do the executives who need to know (not that many of them do). Together, we can figure me out.
In the meantime, there are always cat pictures.