043: OUTSOURCING

<<First ………. <Prev ………. [Archive] ………. [Map] ………. Next> ………. Last>>

“Artificial intelligences aren’t truly intel – ”

“Nope!” Tal raises a finger. “Artificial brains aren’t truly intelligent. Circuits can’t do what neurons can, no matter how you design or program them. But you know what can do the job of neurons? Neurons. You guys have seen those robots with the mouse brains, right?”

“They’re cute, but useless,” Lina says. “Nobody’s going to put a pile of mouse brains in charge of an interstellar ship, no matter how much actual hardware is around it. That’s pointless. They can’t do the job.”

“Also, they can’t survive for thirty five years,” the Friend says.

“And they are not on ship,” Denish adds. “I have looked at all schematics and checked systems with my own eyes and hands. No giant tanks of brain matter.”

“And I’ve inventories all storage areas,” Tinera adds. “None of the artificial wombs are in use, obviously. Nothing weird in the lab or medbays that the doctors didn’t put there.”

“There might be parts of the ship that I’ve missed in maintenance so far,” Adin says, “but there definitely aren’t any tanks of brain matter lying around.”

“Oh, but there are!” Tal leans forward over the table. “There are nearly four thousand tanks of brain matter lying around.”

We stare at kem.

We stare at each other.

We stare into space.

“That’s not… possible,” I mumble. I look to the doctors. “Is it?”

“I, I would have said no, it isn’t, but…” Lina looks at the Friend. “I mean, synnerves transmit and receive information. Could you… offload some decisions to… I mean, I don’t know how computers work, but…”

“Forget computers, can brains do that?”

“Oh, yes.” the Friend nods. “You see it with traumatic brain injuries all the time. Even in a healthy brain, bits are just off doing their own thing, running their own calculations, and then at the end a bunch of the data is gathered together and the brain pretends it was acting at one unit the whole time. A cerebral stimulator is supposed to stimulate hallucinations – dreams – in the comatose patient; that’s its entire job. If there was, theoretically, a way for it to get coherent data back…”

“Then it could ‘learn’,” Lina murmurs, “to send specific things, converse through trial and error… could that work? Use machine learning to establish a way to give and receive information from the sleeping brain, and have the brain learn to respond?”

“No idea,” the Friend says. “How would you even go about figuring out if that was possible? Figuring out how to do it?”

“Run the experiment, I suppose.”

“Whoah, whoah. Hang on.” Tinera throws up her hands. “What, exactly, are you guys talking about here? Are you suggesting that the AI of this ship is some kind of, of human brain hivemind existing in the minds of our sleeping colonists?”

“Oh, no, that wouldn’t be possible,” Lina says.

“Right, good, because it sounded like – ”

“What Tal is suggesting is that our perfectly standard ship’s AI can, through as yet undetermined means, interface with more specificity with human brains than intended, via the cerebral stimulators. And that it is presumably offloading some of its tasks into human dreams.”

Tinera glares. “Oh. That’s so much more sensible.”

“It makes perfect sense, when you think about it,” the Friend says. “Conceptually. People tend to remember dreams as weird and wacky, but most of the dreams we have are in fact very normal. They’re simulations of going about our day, practicing mundane tasks. And a dreaming brain will accept and process odd kinds of information without question. It will sometimes reach ridiculous and nonsensical conclusions, but if you put your data through a whole lot of brains simultaneously, and collate the data with your own included, the most egregious errors should even out. It’s one way to give a better simulation of understanding.”

“You’d need a lot more synnerves for that, wouldn’t you?” I ask, thinking of the unusually invasive synnerve growth in our bodies.

“You know that scan that shows the synnerves?” Tal asks. “How well can you do it?”

“What do you mean?”

“Could you get a really clear picture of just the brain?”

Lina rubs her chin. “The machine isn’t designed for that, but we could try.”

The Friend nods. “We should make that a priority. We have to know what that thing did to our brains.”

Modifying the machines and taking scans takes Lina, the Friend, and Denish the better part of a day, and I’m not really involved except for having my brain scanned. I spend my time trying to freak out as little as possible about things I can’t control.

We gather in the medbay before bed. Lina paces back and forth, hands fluttering; the Friend stands stoic at the computer terminal.

“Okay,” Lina says, breathing a little too fast to be calm, “here’s what we have so far. Keep in mind that this is mostly conjecture. We’re viewing this synnerve thing as a sort of two-stage infection, infecting everyone aboard the Courageous. Stage one is an unusually extensive synnerve growth stimulated by the cerebral stimulator. So far as our scans show, the synnerves in stage one – which we all have, obviously – are more prevalent than expected, but don’t appear to be doing anything particularly sinister. They’re just there. This infection drops revival viability to a bit over half, but carriers of the DIVR-32 geneset are immune.”

“Do we know why?” I ask.

Lina shrugs. “Could be any number of reasons. Better blood pressure and temperature regulation are both good things to have to survive a coma, especially if these synnerves are unintentionally affecting regulatory systems. My money’s on the increased resistance to anoxic brain damage, personally. Without knowing specifically what’s dropping the viability, it’s impossible to be sure, but there are many potential dangers that DIVRs have resistance to. We also noticed that the DIVRs scanned had slightly less invasive synnerves within the brain itself, especially in the region of the hippocampus, but we only have scans from seven people, so that may or many not be a real difference.”

“Stage two infection,” the Friend continues, “involves a massive increase in the growth of synnerves. We can’t scan the synnerves in the dead since they can’t metabolise the tracers, but we compared the mass of nerves attached to the cranial ports from our stage 2 dead and even when cleaned of foreign material, they’re… bulky. This friend estimates that they have at least ten times the concentration of synnerves that stage ones do, by comparing the cranial ports to our scans. Also, knowing what to look for, this friend reexamined their skulls and it believes we have an answer to their fragility. It’s not about the growth and maintenance of the bone tissue at all – the ports simply tried to grow so many synnerves that they grow through and crack the bone.”

“Gross,” Tinera says.

“When you say we’re, uh, ‘infected…’” Adin says.

“There’s nothing for any of us to worry about.” Lina grimaces. “That I know of. Firstly, all of us except for the captain have been through the electrostatic shield, which we know killed the synnerves in your artificial foot, so we can assume that the synnerves are completely unviable in everyone except Aspen. Second, even if they weren’t, they’re not dangerous. There are myriad dangers to synthetic nerves being laced randomly through the body, but if you haven’t experienced symptoms of anything strange yet, I wouldn’t worry. With our cerebral stimulators detached, they can’t grow any further. Even if you stuck another stimulator in, they wouldn’t do anything; it would grow a new network of nerves. That’s why repeated chronostasis is so dangerous; the nerves grown uniquely each time and can only communicate with their stimulator. So, yes, we’re ‘infected’ in that these synnerves are in our bodies, but they’re not going to grow and do anything new.”

“Unlike the people still in chronostasis,” I murmur, “who are at risk of developing stage two.” Should I take a trip through the electrostatic shield? Just in case? It probably isn’t worth the risk, but the synnerves are really starting to creep me out.

“Presumably,” the Friend agrees. “We assume that stage two patients are the ones whose dreams are being outsourced to by the AI, but we have no idea what stimulates the progression. Whether the AI chooses it, whether it’s something that just flares up when the body can’t resist any more, whether there’s some other trigger. We don’t know how to predict it or how to protect people, other than waking them up.”

“Stage two only happens in CR5, and in CR1,” Denish points out. “Must be environmental. CR1 now gone, so we only need worry about CR5.”

The Friend nods. “Probably.”

“So waking more people up from CR5 might increase their chances of survival, by getting to them before the AI does. Or, depending on how exactly stage two is triggered, it might just be waking people up, putting strain on the ship, and putting everyone in danger for no reason. We basically have no idea whether waking people up is helpful or not.”

“Better question,” Tinera says. “What about the people already in stage two? What do we… do… with them?”

We all look at each other in grim silence for a bit.

“They’re dreaming, right?” I ask. “They can’t wake up, but they’re essentially dreaming right now?”

“That’s our hypothesis,” Lina says.

“Are they in pain?”

“There’s no way to tell.”

“So… what are we doing with these people?” Tinera asks. “Are we just gonna let the computer do this to them?”

“The alternative is killing them,” I point out.

“Reimann seemed to think that was the right call.”

“Yeah, well, I’m not sure how much I trust Reimann with – ”

“The people who put this together did. They trusted his judgement enough to put him in charge. And we don’t know how he found out about all this, or how much he knew. I’m not saying we should start axe murdering, I’m just saying that we need to be braced for the possibility that he knew something that we don’t. Maybe he just found this personally unacceptable and snapped… or maybe he knew more than us, and he knew that destroying this mind farm thing was the right call. We need to keep ourselves open to the possibility that this might pan out to be even worse than it sounds.”

<<First ………. <Prev ………. [Archive] ………. [Map] ………. Next> ………. Last>>