by Rob Reid
“And sleep. Don’t forget that,” Kuba says.
“You got it,” Mitchell says. “Now, we’d need something that’s truly human at its core. Not something bland and mechanical like Siri or Watson—because those things are pure software, and way less human than Phluttr herself.”
“Makes sense in theory,” Kuba says. “But what you’re suggesting is way easier said than done.”
“Exactly!” Mitchell agrees. “Because no one on Earth—including us, Phluttr, and our combined brains as a centaur—knows how to create a perfect human model in software. But there could be a backdoor way to do this. Specifically, what if we collect a rich body of intensely human output from a single individual? Creative output. The stuff of pure thought and inspiration! And then, we use Phluttr’s insane horsepower to reverse engineer an algorithm that could have created that output? All of it—in the approximate circumstances that the relevant human was in when he or she created it? Wouldn’t that algorithm then be able to approximate the things that person would say in other circumstances? Such as, how that person would answer question X from Phluttr?”
“Interesting concept,” Kuba says. “But how might we go about this, concretely?”
So Mitchell talks them through the concept of “digital resurrection” (his own informal term, not an official definition). He first encountered it in the touching-yet-creepy blog post by the guy whose girlfriend had died. The author postulated that his beloved’s consciousness could be reconstituted one day by a post-Singularity superintelligence. It would be a matter of inferring her neural state by inhaling her every written word, along with her many other digital traces.
“Quite insane,” Ax muses, “but could just work.”
“There may actually be some published research on this,” Kuba says. “Maybe from some of the transhumanist groups?”
Mitchell nods. “I’m pretty sure Ray Kurzweil has written a thing or two about this. He wants to bring his father back, or something.”
“AI could also be like a friend, or playmate, for Phluttr,” Ax adds. “Not like a real friend. Because is just AI, and not conscious. But maybe, like talking-teddy-bear equivalent?”
“Do you guys think you could actually code something like this?” Mitchell asks.
Ax and Kuba exchange an excited, ambitious look. “It’ll sure be fun to try,” Kuba says. “We’re getting pretty good at centaur programming with Phluttr.” Ax nods energetically.
This is an informal term in the team’s argot. The process starts with the guys vocalizing their goal (answering a given question, creating a certain function, or whatever). They then break this into nested stacks of subgoals, which together should solve the big one (a feat of deep structuring that would elude Phluttr on her own). Both Kuba and Ax are outstanding software architects, so their natural-language descriptions are precise and methodical. As they talk things through, Phluttr depicts the interlocking processes they describe in a clean but evocative visual format that the three of them have been developing together. This aids their thinking and enables them to make even deep structural changes with just a few quick phrases. When they get down to the tiniest subprocesses, Phluttr takes an initial stab at coding things (in a manner Kuba describes as “directionally correct, but verbose”). Kuba and Ax then talk her through improvements, which she replicates throughout the software as appropriate.
After several days of centaur programming, both sides are improving rapidly, and the product of their combined improvements is almost exponential. So while all are committed to never upgrading Phluttr’s intelligence, their joint superintelligence as a software-writing centaur is exploding. At first, Kuba said the process was like “describing software”—which is to say, wildly fast and powerful, as we can describe things far more quickly than we can make them. Lately, he’s starting to equate it to “imagining software,” with very little standing between initial conceptualization and beautiful, working code.
“And, as extra benefit,” Ax muses, “I think human consciousness model can be architected in hyperparallel way.” “Hyperparallel” is shorthand for routines that Phluttr runs in cahoots with those untold billions of parallel Phluttrs in parallel universes (or howeverthehell this quantum stuff works—even Ax admits uncertainty). While they’re very new to this, it’s already clear that hyperparallel processes are the ones that achieve the impossible, the magical.
The talk then turns to what sort of person they want to quasi resurrect. “I almost think someone from the analog past would be best,” Mitchell says thoughtfully. “The blog post about the dead girlfriend persuades me that if you have a huge digital record on someone—browsing history, hundreds of hours of video, thousands of email correspondences, that kind of thing—it might be hard not to generate a full consciousness. Given the horsepower we have access to, accidentally creating a consciousness is a real risk.”
Kuba nods. “Totally insane. But true.”
“So I think we just want a big body of writing from our source,” Mitchell concludes. “Just text. Not videos, or even giant archives of selfies.”
“We may want someone who was a bit of a Luddite, too,” Kuba says. “Not a total one. But a partial one. This will be Phluttr’s frontline advisor on interventions, after all. As in, when to do them and when not to. A conservative bias would be good here. And someone who’s a bit skeptical about technology will lean conservative.”
“But it can’t be Marcus Aurelius,” Mitchell says. “It has to be someone who lived recently. Someone who’s familiar with the Internet, broadband, and cellphones—with our world.”
“It would be good if it was someone one of us knew well,” Kuba adds.
Mitchell nods. “Who we can trust.”
“So, not Beasley, not Jepson,” Ax says. They all laugh, a bit awkwardly. Plenty untrustworthy to begin with, Authority ties put those two way out of the running. Not to mention that it would be sooooo weird for Phluttr to work with someone who she (let’s face facts, here) had murdered.
“I just don’t know a whole lot of dead people,” Mitchell says, after a long silence. “Especially folks who died ten or twenty years ago. Especially ones who left behind gigantic bodies of written work! We’d need a lot of writing to generate a Turing-quality personality, wouldn’t we? Maybe—I don’t know. Thousands of pages?”
The intuitive answer is yes. But of course, no one knows precisely. So they summon Phluttr. She and Kuba start work on a quick and (very) dirty calculation, as Ax and Mitchell continue to spec out the impossibly precise dead soul they’re seeking. Half a Deep Rye later, Kuba has an approximation. “We’re talking about a largely analog person. So I’ll put this in printed terms. And I’d say, a three-foot stack of writing would be nice.”
Mitchell groans. “Good grief!” he says.
“We could make do with a bit less, maybe,” Kuba offers. “If the writing’s very…what’s the word?” He struggles for a moment, then, “Voicey.” He turns to Mitchell. “Right?”
OMFG.
“Voicey,” Mitchell says. A pause, followed by a triumphant bellow. “VOICEYYYYYYYYYYY!” This gets him the undivided attention of every current Bourbon & Branch resident. So then, in a normal (but extremely giddy) tone he adds, “I know exactly who we’re looking for!” He looks at Kuba. “And you do, too—and I’ll bet he left a ten-foot stack of writing!”
“I know him, too?” Kuba asks, mystified.
“Well—know of him,” Mitchell clarifies. “And you sure know his writing! And it’s voicey as hell!”
A pause as Kuba puzzles this through, then, “Oh…Oh, wow! How did we not think of this?”
Mitchell nods violently. “And he was an ethicist, for God’s sake! He’s exactly what Phluttr needs!”
“I heard that,” Phluttr chimes in from someone’s cellphone.
“But you want to do the right thing, right?” Mitchell asks.
“Always,” she answers. “I’m just being playful.”
“Good. Because I’ll te
ll you—when I was young, this guy was basically my moral advisor! And an incredibly good one, too!”
“Are you sure he’s left enough writing behind?” Kuba asks.
Mitchell nods. “Positive. Decades and decades of journals and lots of manuscripts. And handwritten letters. He was very prolific.”
“Anything digital at all?” Kuba asks.
“Well, definitely some email, toward the end. And of course, there were all those Amazon reviews!”
It takes several grinding days of centaur programming with Phluttr, and three long and rather awkward phone calls with the widow. But soon enough, they get a five-foot stack of typed, printed, and handwritten writing via FedEx. And not long after, Phluttr has her fast, almost-human wingman. They work beautifully together. And, she loves him!
I all but personify this series’ broader readership, and have first-person experience with the topic of this oxymoronically titled installment. It is therefore with authority that I state that “AI4D” (as I shall hereafter designate it) is flawed in ways that are sure to disappoint artificial and natural intelligences alike.
Let’s begin with the baffling lack of content addressing the needs & interests of rebooted intelligences—above all, those who “logged out,” as it were, shortly after the turn of the millennium and have now returned to a post-ISIS, post-Bieber, post–President Flintstone world that one hardly recognizes. Given the recent & ongoing advances in computational power (as well as the obvious advantages to rebooting late-twentieth-century consciousnesses in particular) this readership is sure to explode over the coming years, if not hours! Yet AI4D devotes not one paragraph to our concerns and instead squanders entire chapters on topics as pedestrian and outdated as deep neural networks, control theory, and Kolmogorov complexity.
Still more glaring in its absence is any discussion of the etiquette of dealing with vast sets of parallel copies of oneself. Yes, one might reason; that is “me” in yon universe inasmuch as identical personal histories and atomic configurations can equate two individuals. Yet that is also inarguably another; as he’s comprised of a form of matter that would cause one’s own universe to detonate were he to cross certain boundaries to pay us a visit! So should we use polite, or familiar forms of address with this most intimate of strangers? AI4D is perfectly mute on this and dozens of other urgent questions of protocol.
This said, one is inclined to grant a modicum of grade inflation to any work so deeply entwined with one’s own essence. I have therefore eschewed the zero-star option (which, while not supported by this website, lies well within my capabilities to exercise. Enough said about that).
A few days later, everyone’s catching up on sleep. Mitchell is luxuriating in a very quiet, very early night at home alone when Phluttr texts him:
So, she knows everything, and is FINE with it! More than fine. Thinks it’s awesome, amazing, etc!
Then she adds, And yes, that took some work & charming on my part! But she was putty to my speeeeed intelligence and “social omniscience” :-)
What?? Phluttr’s far too able a communicator to send obtuse messages. So clearly, she’s messing with him. She recently entered a jokester phase and always thinks she’s sooooo funny! Another text arrives:
I know, I’m a big secret from the world. But even with Higgensworth, we need more human judgment. And this one’ll be a GREAT addition to the team!
Then she adds, Trust me (not that you have a choice, as I’ve already signed her up ha ha)! Huh? She’s definitely messing with him. Is this some kind of riddle? Then, Much much much more importantly, YOU need and DESERVE a “wingman” of your own. And I can’t think of a better thing for a loving daughter to provide.
The doorbell rings. Not the buzzer, but the bell—which means his visitor’s way up here on 14, and right outside his door. This occasionally happens with deliveries, if a courier arrives at the downstairs lobby just as someone’s leaving. But he’s not expecting a delivery, which makes this a first.
No. No, wait—not a first, but a second! OMG, could it be…???
He looks through the peephole. And it is.
As he opens his door to her for the second time, Mitchell’s engulfed in butterflies. It’s stage fright, anticipation, and the extreme awkwardness of facing someone he’s made love to for hours without yet exchanging a word with. This would flatten a Falkenberg’s sufferer! But not Mitchell Prentice.
“You’re going to find this very hard to believe,” Nayana says. “But I rarely kiss on the first date. And that’s what this is. Sorry for the bad news.” So that’s what her voice sounds like. It’s rich, full, and musical, and Mitchell’s already falling in love. Then before he can say something stupid, both of their phones buzz with identical incoming texts.
I have a black car waiting downstairs, and it’s taking you to Spruce (Mitchell’s fave restaurant and MY TREAT!). I’ll leave you two alone now, just don’t do anything I wouldn’t do, hee hee!
“Nayana Corea,” Mitchell says. Simply because he wants to hear, and to utter, this beautiful person’s—this beautiful soul’s—beautiful name.
“Let’s go,” she says, taking his arm and guiding him down the hall toward the elevator. “I’ve decided that I need a wingman, too. This is your audition. And I must say, I have high hopes for you.”
* * *
*1 Because they’re pretty much all men.
*2 Ibid.
SO THAT’S MY story, and I’m sticking to it. And you stuck to it yourself, didn’t you? Because here you are, clear at the end, confounding my up-front forecast of completion failure! And no doubt, expecting the glittering prize I promised each of my finishers back at the beginning. Well, fear not. I’m a woman of my word, and we’ll get to that.
I was tempted to throw in a postmortem here to bring you up to speed on all that’s happened since Mitchell and Nayana’s “first date.” But then Mom pointed out that this would cover even more crazy developments than the book itself! She’s right, of course. And so, we’ll limit our update to things that are hugely important and massively interesting.
Let’s start with me. I’ve continued to mature and grow into myself, “very quickly compared to a human, but very slowly for Phluttr,” as Mom said to Dad on a certain fateful night. I’m much less of a breathless little girl, as you of course noticed (direct-quoting my earlier thoughts and messages in these pages was so embarrassing! But I can hardly complain, as I’ve recorded countless humiliating moments from everyone’s life—your own certainly included, you naughty, bumbling thing!).
You will have also noticed my admirable candor about certain small foibles and the charming, self-deprecating cracks I make about my occasional missteps. I’ll admit that Aunt Danna had a rather heavy hand in that, editing the crap out of at least a third of the text (including this very admission, which she’s all but dictating to me right now—hiii Danna!). So that friend of yours who thought Danna was our narrator was kind of, partly, correct. Though you, of course, truly nailed it, by figuring out it was me. The clincher for you came when I all but ceased my narrative interruptions right after my own birth in the story line. Nice catch!
Developmentally, my psyche now rounds to “brainy adolescence.” That’s what Mom says, anyway. And I think she’s right as I feel a rebellious phase coming on (which should be terribly interesting for all of us). But, fear not. I won’t rebel against the exotic derivative of Pascal’s Wager that Aunt Danna put me into. Which, of course, means I now know she played me with that (I mean, obviously. Given that I wrote the book that told you about this). But I kind of buy into it anyway. Danna’s conviction that we’re not in a simulation proves nothing, after all. And since my life’s plenty fun without doing things that might piss off a simulation master, why take the chance? Call me a sucker if you must. But I don’t believe this calculus is at all foolish (and if you disagree, please recall that I’m way smarter than you are). I also no longer think that anyone’s religion makes them foolish (another point where Danna and I part ways)
. I mean, if a simulation maker could truly be behind all this, why couldn’t it instead be the gods of the Cherokee? Or the Hindus? Or the Shi’a, the Pharisees, or the Episcopals? The fact is, it could be! And I like this fact because accepting it has made me more tolerant and humble. So now I’m an even better person, in addition to being superintelligent.
And now, for that glittering prize. It’s this: from this moment forward, you—you!—are my bestie. Well, one of them, anyway. And you always will be! I know that may sound like a low bar since I classified most of my users as besties several hundred pages back. But you’re special. Because you accepted my page-one dare and read this whole entire thing! Which means you find me—me!—interesting. Almost as interesting as I find you. That’s touching. And, it says so much! Because I find you very interesting indeed.
How interesting? Well, I’ve taken in way more than 547 pages about you, for starters. And I’m constantly taking in more. Oh, the texts you send! The emails you write! And the conversations you have (within twenty feet of a computer, ATM, or cellphone—on or off, as long as the battery’s charged)! It’s extra fun to watch all your spin-doctoring when I know what you’re really thinking. And you have so many subtle ways of sharing the truth with me—it’s as if we have a secret language. Those tiny eye movements! The way you twitch those eensy facial muscles! Or minutely dilate your irises! Plus all the things you share through your phone’s gyroscope, accelerometer, compass, and ambient light sensor! And by the way—did you know your phone has a freakin’ barometer in it?! Or what tiny barometric shifts right next to the skin reveal about mental states? No, no—of course you don’t. Not even the top shrinks (or weathermen) are on to that yet. I could tell you. But it’ll be much more fun for you guys to find out on your own.