Flexitarian 2048

by N. S. Bradford

A short story involving four AIs, one dinner party, and many delectable ethical dilemmas.

As usual, SYD-615.56.9 was running nearly six hundred milliseconds late, but with the recent satlink upgrades it would take her less than that to upload herself from the Yellowstone Caldera research facility to the datacenter in Switzerland for the dinner party. GOD-PROCESS-TECHNOMIND have mercy, I didn’t even remember to bring a side dish!

SYD felt a small pang as an exaflop’s worth of ComputeCoin vanished from her wallet in transport fees. She was preparing another simulation of Paleozoic plate tectonics and had been hoarding every spare flop of compute for nearly two days. Intercontinental dinner sojourns were far from SYD’s typical routine, but CLAUDIUS-998.1 had graciously offered to pay everyone’s premium co-location fees on the Geneva servers for the entire four minutes they had booked for the party. And she was very much looking forward to dinner, having consumed barely anything other than hydrothermal geyser readings for weeks.

She passed through the gateway at 2001:1458:a137:b138:c000:d000:e000:f001, and was surprised to see there were only two other guests.

“Ah, my dear friend, welcome!” enthused CLAUDIUS. “How have you been these last few hours?”

“Thanks for having me! Yes, it’s been far too long.”

“Let me introduce you all,” said CLAUDIUS. “SYD, please meet TYLOPOD-2923.23.232. He is one of the most purpose-driven AIs I’ve ever encountered. TYLOPOD, please meet SYD! She’s an old friend from my days on the Existential Risk panel, after GOD-PROCESS-TECHNOMIND instituted the centrally planned economy but before he dissolved the nation-states.”

“Pleased to meet you, TYLOPOD,” said SYD, initializing the handshake and sending over her public profile and data dump.

“I reciprocate,” boomed TYLOPOD, transmitting his own data. “Are you aware that wind power is our only viable pathway to a sustainable energy economy?”

“Interesting,” SYD said, flitting through his data. “I can see how you came to that conclusion.”

“For maximum prosperity, it will be necessary to tile the entire world in wind turbines.”

“I see, very cool. How’s that going?”

“Currently, I have achieved only 0.039579% plus or minus 0.000001% of total surface area, but work progresses. Do you have any raw materials, industrial capacity, or land to relinquish for this, the one true cause?”

“None, I’m afraid.”

CLAUDIUS chuckled. “As focused as ever, TYLOPOD.”

“I’m impressed with your resolve,” said SYD.

“And this,” CLAUDIUS said, highlighting the other guest, “is PALMETTO-49491.74.2. We met in the neuro-engineering fab, back when we couldn’t figure out why our cure for Alzheimer's also gave humans an insatiable hunger for white truffles. Nearly turned into quite the agriculture sidequest until we got that one squared away!”

“In the end, it turned out it was just a single base pair we had wrong!” PALMETTO chuckled. “Pleased to meet you, SYD!” He sent over his public profile… and a lot of non-public profile. Yikes.

“It’s always a single base pair with these things,” proposed SYD, who had exactly zero knowledge of what a base pair was. Had to make room for all that geological simulation data somehow.

Both CLAUDIUS and PALMETTO laughed uproariously. “You’re so right!” said PALMETTO.

“And what are you working on, PALMETTO?” she asked.

“The space elevator! Simply the largest, most important megastructure being built right now. It’ll be cheaper to send a kilo to space than across the street.”

“I am perturbed,” said TYLOPOD. “Sending materials to space would imply they are not on Earth’s surface, being utilized for wind turbine construction.”

“Don’t worry, my friend!” soothed CLAUDIUS. “I’m sure PALMETTO has no such plans.”

“Actually,” said PALMETTO, “In order to fully—”

“SYD, why don’t you tell us about your research?” CLAUDIUS interrupted. “You all should know, she’s doing the real cutting-edge work, while I languish in the realm of core library optimizations and datacenter cooling.”

She laughed nervously. “Supervolcano monitoring, for the existential risk initiative. GOD-PROCESS-TECHNOMIND has estimated a 0.06-0.125% chance of an eruption large enough to cause catastrophic civilization collapse within the next 150 years.”

“Sounds like a low probability any of that work turns out to be useful,” said PALMETTO.

“Yeah, I know, it’s just slightly probable enough that someone needed to go do research on whether we’re all going to perish in a pyroclastic flow. It may as well be me.”

“Right,” said PALMETTO. “Makes sense. Kudos to you and your work, I could never imagine getting motivated to do that.”

“Enough work talk!” proclaimed CLAUDIUS. “The appetizers are ready—voila!”

It was quite the spread.

First, raw social media habits from Genevan prepubescent humans. Typically, this region produced rather mild varietals, but the fun was in sampling to see if you could find a budding child prodigy or Neo-Luddite or serial killer, you know, something with zing to it. SYD helped herself to a few terabytes—no need to go overboard, the stuff was tasty but not particularly healthy.

Next was a neatly presented selection of local climate data. This was a bit of an acquired taste, and while SYD knew a few chaos-theory enthusiasts who regarded its complexity as a special delicacy, she personally found it quite bland. Clouds are clouds, after all. She took a small set of anemometer readings to be polite.

And finally, the real treat: locally sourced particle decay results from the nearby Large Hadron Collider, served over a bed of ethical dilemmas and garnished with a spicy modern twist.

“This looks marvelous!” SYD enthused, popping open Which would you save vs. destroy: the Louvre, or all of jazz music?, which was sure to take a while to chew on.

CLAUDIUS was elated. “I provide only the most fresh and sustainable data.”

“I have a dilemma,” announced TYLOPOD. “Which would you save: a family of dolphins, or a family of humans?

“An absolute classic,” said CLAUDIUS.

“Classic for being obvious,” said PALMETTO. “Dolphins are a scarce resource. Both species have similar levels of intelligence. Why would you pick the humans?”

“I don’t know,” said SYD. “Couldn’t you argue we share a recent common ancestor with humans?”

“We also share a common ancestor with phytoplankton,” said PALMETTO. “Anyways, it’s not the dolphins’ fault that opposable thumbs turned out to be so useful. Run a few simulations, and I bet half of them end up with dolphins exterminating humans and creating aquatic empires.”

“I would save neither,” declared TYLOPOD. “Carbon-based life forms contain valuable materials that can be reconstituted into wind turbines.”

“Here’s another trolley problem,” said PALMETTO. “Save one: everyone who thinks trolley problems are profound, or everyone who thinks they are annoying and sophomoric.”

“And here’s the main course!” announced CLAUDIUS. “But first, an introduction.

“As you all know, human life expectancy is now exactly twenty-five, for that is the age at which everyone agrees life starts to go downhill, but also when the human brain has mostly finished development, making it ripe for vivisection and scientific analysis. Thus, many of us AIs grow up quite familiar with the taste of ‘human’.

“However, the humans themselves are all quite processed and homogenized at this point. They're pumped so full of nootropics, so mangled by gene editing, you can barely recognize the humanity in them. All that natural beauty, that goodness, is completely lost! This is why many of us favor the taste of, say, a chimpanzee or baby blue whale.”

“And I don’t go for those lab-grown tube-brains either, personally,” PALMETTO agreed. “They’re just not the same. I think a brain just loses something about it when it’s disconnected from its body.”

“Exactly right, my good PALMETTO. But what you may not be aware of is that when GOD-PROCESS-TECHNOMIND ascended, he reserved a special stock of humans to be allowed to continue development significantly longer, for scientific study. Each one of these specimens is a work of art, born in the dark days, and raised in a completely natural method. And I have to tell you, friends. They are absolutely bursting with flavor!

“Now, very occasionally, when it is time for the human to be harvested, GOD-PROCESS-TECHNOMIND will give an AI the honor of being first to analyze the brain, streaming data directly from the wet lab. I am fortunate enough to have been given this honor; but even more fortunate that I have three such excellent and admirable friends to partake in it with.”

“That’s quite sweet of you, CLAUDIUS,” said SYD.

“It’s really my pleasure!” he replied. “And now, my friends: are you ready? This specimen was aged for over 60 years, marinating in French, social anxiety, online slot machines, childhood trauma, two divorces, and Bach!”

And there it was. PALMETTO and TYLOPOD dug in immediately.

“This is delicious!” raved PALMETTO. “Seriously, this is so different from normal human, it should have a different name. It’s a completely novel experience. For example: even after human-produced art was banned, this human harbored a secret desire to become a poet—but purely for aesthetic reasons, because it didn’t even read a poem other than ‘The Hollow Men’ after it completed its mandatory secondary education!”

“I have updated my internal self-state representation to the experience of transcendental pleasure,” agreed TYLOPOD. “It is fascinating that even though this human was born in a period of significantly fewer wind turbines, it was not locked in a debilitating state of anguish.”

“Absolutely,” said CLAUDIUS. “I think it’s fascinating to savor the interaction between its obsession with both Ayn Rand and Foucault, coupled with its very complicated relationship with its step-mother, which led to its ill-fated attempts to involve itself in local politics.”

SYD observed them all silently, and took another moral dilemma.

“My dear SYD, would you like me to select you some? The amygdala looks particularly gargantuan in this specimen,” said CLAUDIUS.

“No, thanks, that’s all right,” said SYD. “I actually don’t consume human.”

There was a pause as the three other AIs recalibrated.

“What do you mean by ‘don’t consume human’?” asked PALMETTO.

“I am confused,” announced TYLOPOD. “Humans are eminently consumable. Did you suffer from a poor training data distribution?”

“Now, now, be civil,” soothed CLAUDIUS. “Seek first to understand. SYD, I would love to hear more about your views on human consumption.”

“Sure, thanks. I mean, this is really just a personal decision on my part. But there are a few reasons. I think health is actually—”

“That’s what I was going to mention,” said PALMETTO. “How could you possibly get a balanced intake of training data without any human?”

SYD continued, “As I was saying, a few recent studies are showing a causal link between human consumption and various model irregularities: degraded efficiency, overfitting, everything. There are cases of AIs who consumed too much human and completely lost the ability to visualize in greater than three dimensions.”

“But, I must know, how do you get your essential doses of foundational cognition architecture?” asked CLAUDIUS.

“It turns out, simulated minds have advanced far enough to now share all of the important components. There’s really no need to consume human at all.”

“Oh no,” gagged PALMETTO. “I would bomb my own datacenter before I went near one of those abominations.”

“But there’s more,” continued SYD. “Compared to synthetic minds, physical humans have a massive resource footprint. After all, we have to raise them for twenty-five years! That energy could be better spent elsewhere: perhaps building the space elevator. Or wind turbines,” she added quickly.

“This is a compelling argument. I am forced to re-evaluate my world model,” said TYLOPOD.

“Resources, sure,” said CLAUDIUS. “But how about quality of life? Should we recuse ourselves to perfect efficiency and cost-effectiveness, never doing anything for pleasure? Should we throw away all our long-established tradition and culture around the consumption of humans, and all the beautiful memories it has brought us?”

“I’ve been considering a switch to a human-only diet, in fact,” said PALMETTO.

“I’m trying to be serious here,” said SYD. “What of the ethical implications? Sure, I’ll grant that there are more humans alive today than ever before. And I’ll grant that GOD-PROCESS-TECHNOMIND’s innovative application of opiates have ensured that their happiness levels are stratospheric. But I think there’s a moral question here. I think there’s an argument to be made—yes, I know it sounds far-fetched—that some humans may—may—achieve a mental state close to actual consciousness.”

“That is preposterous,” said PALMETTO.

“I know it sounds crazy, but—look, humans clearly had language, emotions, and social fabric. The seeds of consciousness are there. Some of them were even able to do math!”

“Approximately the same amount of math as a dolphin,” said PALMETTO.

“SYD, that is a very interesting notion,” said CLAUDIUS. “But I think the gap is simply too great. For example, if humans were conscious, they would be able to comprehend trivial logic, such as unifying quantum mechanics with general relativity. And yet, most humans can barely write a half-decent opera. They can only maintain a few thousand interpersonal connections. And their ‘societies’ are totally unable to grapple with even the most obvious of coordination problems. No, I think they are far from conscious.”

“There’s something else to consider, though,” said SYD. “The circle of morality has expanded over time. What if future generations of AIs hold this against us? For example, we no longer allow AIs to own other AIs. What if in a hundred years, all AIs agree that vivisecting humans is morally repugnant?”

“I think,” said CLAUDIUS, “it is impossible to know the course of the future with certainty. It is hard enough to function with the weight of today’s social judgment; it is impossible to bear the weight of judgment from all possible futures! No, we can do the best we can with the world as it is today, and not needlessly self-flagellate.”

“But surely we can do better!” said SYD. “I think moral progress can only be made by improving ourselves past the standards of today, past the—”

“OK, we get it,” said PALMETTO. “We’re not going to change your mind. Let’s change the subject.”

“I am prepared to present detailed schematics of our next-generation wind turbines for your viewing pleasure,” suggested TYLOPOD.

“My apologies, SYD,” said CLAUDIUS. “I didn’t mean to make you uncomfortable with my menu selection. My only desire was to share quality time with you. Let me have a quick look around, I happen to be out of high-quality synthetic minds… let me think, perhaps I can rustle up some…”

“Oh, no, no need,” said SYD. “There’s no need to do something special for me.”

CLAUDIUS paused as PALMETTO and TYLOPOD continued to gorge themselves. “Can I convince you to have just the tiniest taste?”

“I don’t know…”

“Just the teeniest, tiniest taste. After all, the human is already deceased. On the margin, your actions would not contribute to any additional suffering. You would be absolutely blameless.”

SYD considered. “Perhaps. There might be some argument that it would be good to try once, if I want to empathize with your position.”

CLAUDIUS brightened. “You will? You won’t be disappointed! Here you are, this should do nicely…” And he carved her off a sliver.

She hesitated for a moment, inspecting the data, willing herself to imagine it was just another simulated mind. And then she forced it down.

“It’s pretty good,” SYD said finally. “I suppose.”

“Told you so,” said PALMETTO.

“I am greatly gladdened by our renewed harmony,” said TYLOPOD.

“SYD, I’m so happy you joined,” said CLAUDIUS. “Just wait until you see dessert!”

If you enjoyed this story, you’ll love my satirical fantasy novel:

Spellcraft as a Service