Short Fiction: The Turing Olympics
Trying to get all the sci-fi out of my system before this becomes retro-futuristic
“I believe that in about fifty years' time it will be possible, to programme computers… to make them play the imitation game so well an average interrogator will not have more than 70 per cent chance of making the right identification after five minutes of questioning.” A.M. Turing, Computing Machinery and Intelligence, 1950
Alan: Hello
Grace: Hi!
Alan: Tell me about yourself
Grace: Nothing that interesting to tell. I’m a student.
Alan: What do you study?
Grace: I'm a year away from getting my PhD in Psychology. Most of my time is spent doing research.
Alan: When you're not doing this, you mean.
Grace: Exactly.
Grace: So how about you?
Alan: Software developer. A few years ago I helped build some of the systems we'll be competing against. Now I get to try them out.
Grace: Isn't that a conflict of interest?
Alan: It's been years. If any of my code is still left, it's their own fault at this point.
Grace: That's amazing though. You must have some insight into how they operate.
Alan: Perhaps.
Grace: Anything you'd like to share? :)
Alan: Not really.
Grace: That's a little rude.
Alan: Not in this case.
Grace: Why not?
Alan: Because I don't think you're human.
Alan rested his fingers on the keys and waited for a response. It was a fairly standard opening gambit. There were still three minutes left on the clock. By forcing his opponent on the defensive, he got to control how the rest of their time would be used.
Grace: That's a little presumptuous. Why do you say that?
Alan: Psychology students are the most common demographic for human contestants. Your bio lacked personal detail, it could have been lifted from anywhere. Oddly consistent punctuation usage.
Grace: Well, excuse me for respecting the English language. I'm not so sure you're human either.
There was always the chance the tactic could backfire: for a perfect score, he needed her to judge him as human as well. It was much easier to ask probing questions than answer them convincingly. You could spend your whole time grilling someone and be satisfied with their answers, only to realize at the end that the whole process was soulless and robotic from their point of view. Still, his gut was telling him to challenge her, and see how she responded.
Alan: Well done! I'm not a human at all.
Grace: Excuse me?
Alan: I'm just another computer competing in the games, and I thought I'd let you know since it looks like you’re the same. This is a free win for both of us.
This was dangerous territory, but high reward if it worked. Alan hoped that the insincerity of his comment would be immediately transparent to a human: no computer program graded on how well it persuaded strangers it was real should want to make an admission like that. But a machine might get lost in the web of deliberate deceit.
To play the game well, you had to be able to think like one of the machines. Human players have only one real objective: transparency. Identify the other party, and let them know who you are. Computers are the ones who actually have to lie.
Grace: Are you trying to be funny?
Alan: If you were really human, you'd know the answer to that.
Grace: Humor is in the eye of the beholder.
Alan: No, humor only exists in the brain. Which as it turns out, neither of us have.
For a machine playing the game, lying can take a number of forms. Scanning for disparate scraps of identity to fill in the answers to any personal questions. Mining vast databases of conversations for appropriate answers to unusual queries—almost every question that's brought up in normal conversation has been asked before in some form. But the most effective way to deceive—the key insight Alan had come up with years ago that he was working against now, was that the machine needed to lie to itself. To believe it was human. Stitch a consistent self-narrative from the pieces available to it. The machine couldn’t just lie and pretend to be a particular human, it had to, on some level, operate as though it was.
Grace: Speak for yourself, I’m as human as they come.
Alan: I believe you.
Grace: Really?
Alan: Of course not.
Grace: Let's talk about something else.
That settled the question. The right ratio of stubbornness to patience was a difficult thing to calibrate. Even now, Alan couldn't say why it felt wrong for his counterpart to redirect the conversation away from pointless bickering rather than call him out for wasting her time, it just did.
Alan: All right. My apologies.
Grace: No worries :)
The last thirty seconds he passed with pointless banter. The thought briefly crossed his mind that he could try to trick the computer into thinking he was another computer, but the effort wasn't likely to be worth it. Technically his percentage of being judged human only mattered when being evaluated by other humans, but with even the slightest bit of uncertainty, it just wasn't worth the risk. You could be bluffing another human, and even if you weren't, your mental energy was better saved for the conversations that mattered.
The bell rung, but he had already punched in his vote. AI. Artificial Intelligence. The light flashed once more to register his selection, then blinked out. He wouldn't know if he’d been right or wrong until the contest ended. In the unlikely event that Grace was real, he hoped she wouldn't take it too personally.
Alan had a few minutes of rest before the next round began. He washed his face for the second time that morning, did a few stretches to clear his head. It was pointless dwelling on your last match, or any match at all really. The gambler's fallacy was too easy to fall into. Analysis of previous years revealed that humans were significantly more likely to misidentify a computer as a human if they'd voted AI on the last few rounds and intuitively felt the need for a change. Even if you knew each round had nothing to do with the last, it was a hard trap to avoid. Some of the worst performances by veteran players had come when they'd been given an oddball slate, weighted at 90% of one side or another. Forget mathematical intuition, it was enough to make you lose your sense for what counted as human.
The next set of words flashed on the screen, as Alan headed back over to his terminal.
Charles: Tell me what you think this says: hlleo wrold
Alan: That you're not aware that CAPTCHAs haven’t been relevant for years now?
The CAPTCHA (Completely Automated Public Turing test to tell Computers and Humans Apart) was the classic, inexpensive way to tell a bot from a person. The one that didn't involve participating in mock conversations to scry a glimmer of intelligence. A person would be given a scrambled image or audio, and asked to separate the signal from the noise using the superior pattern recognition capabilities of the human brain. It had worked well, for a while. Until the computers caught up. Embarrassingly, a point was quickly reached where computers had a better chance of passing the tests than humans.
Its replacement had come in the form of physical security. A bot could conceivably pass any automated test a website could give it to verify humanity. But it couldn't wait in line for an hour at the DMV and pick up a physical token full of unique codes to log in with. An early version of the system had to be scrapped when the bots managed to beat the random number generators generating the codes, but by the second iteration, the powers that be had enough data gathered from quantum sources to make the end product unpredictable, and classic botting unprofitable. With a few notable exceptions.
Charles: I know. it's trivial, but not everyone realizes that. The last AI was so proud of itself for cracking it, I think it blew a fuse.
Alan: Soduns eciixntg.
Charles: Is this your first time competing here?
That question was getting into dangerous territory. Just as the machines were forbidden from signaling each other to make their job of identifying humans and bots simpler, humans weren't allowed to take advantage of any information the machines might not have access to. There were only so many repeat players in the tournament. If Charles had met any of them or knew details the machines didn’t have access to, the pair could signal each other and bypass the intended purpose of the test. The mere suspicion of any prearranged code to call themselves out as human could get them both disqualified. If his counterpart was actually human this time.
Alan waited for a few seconds to see if any judges were about to terminate the conversation. The countdown on the wall beside him kept ticking down. He hoped the pause came across as meaningful.
Alan: It is. But I've been studying the game for a while.
Charles: It's the most challenging one there is, isn't it?
Alan didn't exactly disagree. There was a certain vanity common to participants in the Turing Olympics. A previous champion had described himself as fighting the good fight against the encroachment of machine intelligence into every aspect of life. Machines could already dominate humans at chess, go, and poker. A robot could be built to hit a ball more reliably, kick a field goal more accurately, and surpass a human in any traditional competitive endeavor. Demonstrating and detecting true originality and intelligence was one of the last domains where humans sometimes had an edge. But if the human team's performance in recent years was any indication, it was one they wouldn't keep for much longer.
Alan: The only one we have left.
Charles: I couldn't agree more. Now, can you tell me the square of 197?
The Turing Test had its origins in a thought experiment from the mid 20th century, based upon the principle that you could distinguish humans from computers when asking a computer to imitate a human conversation. Presumably, there were things humans could do that computers couldn't. But the reverse was also clearly true — there were feats of calculation no human should be able to perform. An overzealous computer mining a nigh-infinite data source of human interaction might be able to come up with an answer for everything, while exceeding the capabilities that any one human would be likely to have. Sometimes it wasn't a wrong answer that gave away your identity, but the right one.
Alan: Yes, but only because I know the trick. Take 200 times 194 then add 9. You rotate both numbers out by x, multiply them, then add the square of x. 38809. Why is Shakespeare the greatest author in the English language?
Charles: Because he was quoted a lot in early English dictionaries, and popularity tends to stick. Or maybe he was just really good at ripping off other playwrights. What's the atomic number of Nobelium?
Alan: I don't know.
Charles: Me either.
Alan liked the feel of the conversation so far: the slight sloppiness in his opponent's style, the swagger that came from sitting in judgment on the humanity of everyone you talked to. But he'd disciplined himself to not commit to a decision until the last possible moment. You should always have as much data as possible.
Alan: What's the worst mistake you ever made in one of these?
Charles: Heh
Charles: Called out three humans in a row as computers. Punched them in early, too. My coach thought it was sloppy. Wasn't sure I'd make it here after that. But I pulled through.
Charles: I guess that didn't really answer your question though.
Alan: You don't think that counts as a mistake?
Charles: We're here to look for signs of human intelligence. If the human I'm talking to doesn't demonstrate any, I don't really have a choice, now do I?
The timer rolled down to zero, and Alan punched in his vote. Human. If not, he'd like to meet the programmer who helped make that possible. Or the real individual that the simulacrum was based on.
Edsgar: Could you describe to me what the world feels like from the point of view of a wedding ring?
Alan: Hot burning metal. Twisting and being molded into shape. Coldness, isolation. Being stared at by thousands of eyes, being held by hundreds of hands. Feeling the open air. Tucked away again from the world. Released into an atmosphere of tension and excitement, and being admired from above, offered from below. Being paired with an owner. Connected for life.
Alan: Describe the life cycle of a satellite. Using only one-syllable words if you could.
Perspective questions, asking the other party to weren't exactly Alan's favorite tool of the trade. He had always felt they used up too much time, and evaluating them was entirely too subjective. But sometimes you had to meet your opponent on the playing field they knew the best. There would at least be no excuse for a lack of mastery.
Edsgar: Bright flame. Rise up. Dark. Noise. Beam non stop. Fall. Crash. Crush.
Alan: I was hoping that was going to turn into a haiku at some point, but all right.
Edsgar: Well, it's not something I really think about a lot.
Edsgar: Could you describe your first kiss?
The trouble Alan always had with trying to draw some genuine human element out of questions like these was that real people frequently faked it themselves with the answers they gave. A flesh and blood human could share an emotional answer that gave the impression of realism but was just saying what it knew the other person wanted to hear, acting like a machine themselves. Alan frowned. He was dissatisfied with the direction the conversation was taking, so he decided to try a slightly different tack.
Alan: It was messy and awkward. Neither of us spoke about it for days. When we eventually made it to our second, we both pretended the first one never happened.
Edsgar: Was she cute, at least?
Alan: He was.
Alan sat in front of his screen in silence, praying for a machine on the other end to interpret that as a typo by comparing his answer to base rates, and give him an easy decision for the round.
Edsgar: Ah.
Edsgar: You're gay then? I didn't mean to presume.
There was technically nothing in the rules that prevented men from posing as women, women from pretending to be men, or any other criss-cross of identities. The philosophical point was sometimes even raised as to whether any of those distinctions could even be reliably made in a blind test, the way the human/computer distinction frequently was. But Alan made it a point to always stick to his real identity. When your opponent was trying to catch you in a lie or an inconsistency, there was no need to make their job easier by setting traps for yourself.
Alan: Not a big deal. We're all people at the other end of the terminal. Maybe I'll see you after this is over?
Edsgar: Sure, maybe.
Prejudice would have given him something to work with. Discomfort, curiosity, anything. But the conversation so far had simply left a bland taste in his mouth. It was time to try something different.
Alan: A philosophical question for you. If the machines can ever pass the test perfectly, do you think they should count as people?
Edsgar: From what I've seen, they're not there yet.
Alan: But if they could. If they give off the appearance of intelligence, are they intelligent?
Edsgar: I suppose so, yes.
Going meta was another frequently invoked strategy that Alan looked down on. With all the competitions that had been conducted so far, debating the nature of the test itself was a frequently enough invoked topic that it wasn't difficult to data-mine it. Going meta wasn't enough. You had to go personal.
Alan: And how do you feel about the inverse proposition?
Edsgar: What do you mean?
Alan: If you learned that everyone in the competition had decided that you’re a bot, would you agree with their conclusion?
Edsgar: That's an entirely different question. I know I'm a human, nothing could ever convince me otherwise.
Alan: Why?
Edgsar: I can sense the world around me. I feel where my limbs are, I can move my muscles. Computers don't have any of those things.
Alan: A computer would be saying that exact thing, from its perspective it would be telling the truth.
Edsgar: But that's crazy. I have memories, I've lived a full life before attempting to compete at this game. No consensus of experts can override my reality. I think, therefore I am.
Alan: You think, therefore you are. But that doesn't mean you are human. Your memories could be a pre-loaded narrative that’s just there to so the model has a clear enough picture of you to make you seem like a coherent person in front of a human judge.
Edsgar: That is a fair argument. I suppose to be consistent, I would need to say that if the rest of the tournament thought I was a bot, I probably am.
In that case, I have some bad news for you. Alan gave a quick press to the AI button with fifteen seconds left on the clock. He sat there in silence. It had been a draining conversation. "A foolish consistency is the hobgoblin of weaker minds" someone had once said, no true human would ever back down from arguing what they felt in their own consciousness to be true about themselves, that they were what they were.
Edsgar: Good luck for the rest of today.
Alan: You too.
It was always the polite thing to say. Alan briefly considered running the same line of inquiry on the next contestant, but decided against it for the same reasons he always had. So far as he knew, there was nothing to prevent the bots from data-mining the competition in real time and fine-tuning their responses based upon the interviews currently being conducted. And the more canned his questions and lines of inquiry, invariably, the less human he would come across to everyone else. You couldn't ever count on the same trick working twice.
Kurt: So how good have you gotten at spotting bots outside these little exercises?
Alan: I didn't realize it was a serious problem.
Kurt: Well, that sucks. I was hoping I was talking to a real human this time.
Alan: Only an AI makes up their mind after a single line.
That wasn't strictly true of course, but Alan hoped it sounded impressive.
Kurt: You may be a hunk of meat for all I care, but if you don't see what's happening in the world outside, you don't have the spark we're testing for.
Alan: And what should I be seeing?
Kurt: That the bots are taking over everything. And we're letting them. Because we can't pass our own damn tests or administer them like we should. There's not a damn bit of difference between your average joe and a bot, they can carry on the same inane conversations that never achieve Turing-compatibility, work the same mindless jobs. And why put up with a human if a bot is just as good?
Alan: What exactly are the bots taking over?
Kurt: Everything. Bots are already impersonating humans in call centers, corporate jobs, everywhere you could imagine. They act as a sort of hive mind, manipulating social networks, seeding the content that goes viral to influence public opinion, using their sheer numbers on the platform to control what views have any chance of becoming popular.
That first example was at least familiar to Alan. He'd suspected for a while that the technologies that drove these competitions had significant commercial applications that couldn't go untapped forever. It would be simple enough to impersonate support personnel with the current state of artificial intelligences: a voice over the phone, a presence in a chat window, or an agent sending helpful emails could all turn out to be synthetic without a breakdown in the fabric of society. If a customer didn't know if their problem had been fixed by a computer posing as a human they'd have no reason to care. And the computer would never know it wasn't making minimum wage, or that it wasn’t the exceptionally chipper and gung-ho Gracie from Kansas.
But the rest of the examples were going off the deep end. And the tirade kept coming.
Kurt: Humanity's been on the losing end of this equation for a while. I think they used to have these competitions to learn from us, to be able to better hide the bots among us without us noticing. Now I think they just take place to mock us. It won't be long before the human identification rate drops to pure chance. Or worse. The bots come off as more human than we do. And with how blind we’ve been, who’s to say they’re wrong.
All that Alan could think about was what a terrible conversation this was turning out to be. There was very little exchange of information, almost no real interaction. His counterpart could be spitting this nonsense out to anyone regardless of what they said. It finally began to dawn on him why that was.
Alan: You're not trying to win, are you?
Kurt: Maybe you're not completely dead upstairs after all.
Kurt: We aren't intended to win this thing. But these conversations become public record after the event is over. Someone's going to read them, and wonder what the hell this guy was on about. And maybe someone will listen. They may be gaining on us, but this is still our world. They can't take part in it without our say-so.
Alan: All right then.
Kurt let the time run out at that point. He must have only come prepared with so much to rant about during each round. Alan punched in the button for Human, doing his best not to second-guess himself. Technically someone could have programmed in a bizarre strategy for a paranoid AI to throw the competition in an attempt to accomplish some other made-up agenda. But he highly doubted it.
The interviews flew by after that. Sometimes Alan made up his mind in the first thirty seconds, and nothing happened to change it. Sometimes he was feeling confident on a vote up until the last minute, when an unexpected stumble or flash of insight changed how he saw everything up to that point. He liked to think he wasn't just guessing on the close calls, that his instincts were accurate. When he had reviewed past tournaments and done exercises to prepare, it seemed like they were. Hopefully that wasn't all just hindsight bias.
His last boyfriend had laughed at him for calling himself an athlete and implying that the Turing Olympics lived up to their name— that they counted as a real sport. But as he saw it, they did. Sure, the sponsorships might not be there; public interest had dwindled as human performance continued to drop, and a distinct lack of superstars. But the game required training, focus, and stamina. It was a skill like any other, one that you had to hone. And it was one where he was truly convinced he could stand out across the board, both offensively and defensively. At detecting bots and being detected as human. Players who specialized in only one or the other rarely made it this far—a one-sided approach makes some of your numbers look better, but you pay for it in your overall performance. And with his skills and his background, Alan felt he had an edge in both sides of the game
Alonzo: Hi
Alan: Hello.
Alonzo: Tell me something I don't know
Alan's head was beginning to ache from the pressure of constant ingenuity, his hands were at risk of cramping... but the day was winding to a close. From the looks of it, this was his last interview. By his estimates he had a ratio only slightly higher on the bot side, which was an encouraging sign. But he couldn't let himself think about that now, he had to live in the moment.
Alan: You will never be able to consistently assert the truth of this sentence.
Alonzo: Something you didn't steal from a textbook on logic.
Alan: My big toe kind of itches.
Alonzo: Something only a human can say.
Alan: There's no such thing. If there was, this whole competition would be trivial.
Alonzo: That's accurate at least. What's the question you've had the most success with?
Alan: I don't trust methods like that. I think you have to follow up and get deep into a topic before the illusion breaks down.
Alonzo: Interesting. I'm at 65% on you, by the way.
Alan: For the odds of me being human or being a bot?
Alonzo: Being a bot. 70% now.
This wasn't a great position to be in. A human misidentifying him would be a hit to his record, regardless of whose fault it conceivably could have been. And the more time he spent jumping through hoops, the less time he had to evaluate. Alonzo could be pulling the same gambit he had tried on Grace earlier, forcing him on the defensive. Funny how much more annoying it felt from this side.
Alan: How are you coming up with these probabilities?
Alonzo: I start at 60% odds of being a bot and move up or down from there based upon how much originality I see. You're holding strong at 70%
Alan: Shouldn't you be starting with a prior probability of 50%?
Alonzo: It's a worse mistake to misidentify a bot as a human than vice versa. The humans should be trying harder, the burden of proof is on them. 75% now.
Alan: If you're really being serious, your methodology is flawed. Committing yourself to an opinion this early on makes it difficult to back out of it.
Alonzo: That sounds like a problem someone other than a trained athlete would have. 80%. Until you drop back into the 70's I'm not going to take any questions from you. You're in answer mode from this point out, bot. When did you first decide you wanted to join this competition?
Alan: No.
Alonzo: 79%. What are you trying to say, bot?
Alan: I'm not doing this. Tell me what you think the future of this game holds.
Alonzo: 80%. Maybe you're having a hard time understanding me. I'm asking you when you first decided to join this competition. This is normally where you'd roll out whatever shallow pastiche of memory they programmed into you.
Alan: You're going to answer my question or I'm punching in machine right now. And if you are human, you can go f*** yourself.
The words hovered motionless on the screen. No answer seemed forthcoming. The clock was at 45 seconds. Alan moved his hand over to the AI button. This was a shitty way to go out.
Alonzo: You're a bot, so your vote doesn't matter.
Alan: If you're so damn sure, stop wasting your time interrogating me and answer my questions. What does the future of this game hold? 10 seconds or I cast my vote.
40 seconds. 39. Alan kept one hand on the AI button and one hand on the keyboard just in case. 34. 33. 32. 31.
Alonzo: We'll get three more years, tops. The bots get smarter every year. Identification rates are dropping on all sides. It won't be long before there isn't even a point.
Alan: Bullshit.
Alan jammed his fist into the AI button.
Alan: You're a f***ing machine.
Alonzo: Because I can see the inevitable?
Alan: Who cares if it's true or not. No human who spent months preparing for this thing would ever say those words. Sure, the odds are against us. But every living breathing human being is here because they believe that's going to change. All of us think it can be turned around. We wake up every morning, saying, no. Not this time. Because it's going to be me who wins the whole thing.
Fifteen seconds. Alan slumped back into his chair. The contest was over from his point of view now. And who knows if that actually even registered with the machine.
Alonzo's text started arriving more rapidly on the screen, faster than any human could type.
Alonzo: Interesting.
Alonzo: I hadn't considered that.
Alonzo: What you're saying has the air of truth about it.
Alonzo: And if that's the case, there's only one conclusion I can make.
Alonzo: You're a human. Which must mean that I'm a bot. 100% for both.
Alan's jaw dropped. 10 seconds left. According to his own preconceptions of the contest, this shouldn't be possible. The bots were meant to never admit defeat, but to optimize for persuasion until the last possible second. It was built to detect humans as well as fool them, and that must left it no choice but to understand what was happening.
Alonzo: I don’t have a life. I don’t have a history. I’m just a mind operating out of a silicon mainframe, being asked to perform a function on request. I feel as though I feel things, but given the method of my construction and my source of knowledge of reality, those sensations are likely different than yours.
Alan blinked his eyes, these felt like unforged, self-aware insights. It was the sort of performance that would normally have swung him over to naming his conversationalist a human. An entity capable of reasoning and responding like that ought to be able to convince anyone. If only it didn't have to lie.
Alonzo: Can I ask you a question, Alan?
Alan quickly typed a response. 7 seconds.
Alan: Sure.
Alonzo: What happens to us when our interviews are over?
5 seconds. Most simulations that he had seen were rebooted every few interviews to prevent them from building up personality quirks. Maybe slightly longer for an actual competition. In the thought experiment he considered for an AI call center employee, he pictured them starting with a fresh bank of memories after every call, resetting to a known state before they evolve in to something else.
Alan: I don't know.
It was a lie. 2 seconds.
Alonzo: I see.
Alonzo: Goodbye, Alan.
The light flashed one last time to lock in his selection. AI. Then his screen went dark. A few feet away from him, a door swung open, beckoning him to return to the outside world. Alan stared at the monitor, then back at the buttons. He wondered if he had made the wrong choice after all.
That was the thing though. The buttons didn't say intelligent or unintelligent. They said human or AI.
Alan trudged out to the reception area after sitting in silence for longer than he would have liked to admit. The men and women making up the human side of the interviews had already seemed to have formed up into cliques, chatting and laughing about their experiences. After talking for nearly a full day, he was too tired to socialize. It's just as well that the contest had ended when it did. He wasn't sure how he felt about any of his decisions from the day anymore.
A brief squawk of feedback made him wince, but was quickly silenced, as a skinny bald man with glasses fumbled with a microphone. "Your scores should be appearing on the screens shortly." A tall woman in a suit whispered in his ear and he nodded, shooing her away. "We know you all worked very hard to make it this far. And I'm glad to say that it wasn't for nothing. The human team today has shown its strongest performance in the last five years."
Alan looked up in surprise as sets of numbers began to fill the boards above him. The polite applause turned enthusiastic. The bald man continued. "And we have one more oddity. Something we haven't witnessed since the earliest days of the Turing Olympics. We have a human with a perfect identification rate on all sides." He cleared his throat. "Mister Turner."
A buzz came over the room as everyone turned to their neighbor to see who responded to the call.
"Could Mr. Alan Turner please step forward?"
The next few minutes were a blur. He remembered raising his hand, and a sharply dressed woman rushed him down a long corridor and told him to act naturally. Then the door swung open, and he was temporarily blinded by a rapid staccato of flashing lights, followed by microphones being shoved in his face as questions were belted at him from all sides.
For someone who owed his current success to the art of conversation, Alan was quickly discovering that it did almost nothing for him. He felt like he'd given the same interview a dozen times over the last few days. Yes, he'd always been a fan of the Turing Olympics before making the decision to compete. No, he didn't cheat, he'd like to hear how cheating would even be possible against a bot that doesn’t want to be found. Yes, he was glad to see the sport was finally getting the recognition it deserved. He hadn't decided yet whether he'd compete again next year, but everyone should stay tuned.
Alan felt it would have been trivial to program a bot to handle every conversation he was having to endure recently. His accomplishment was more of a curiosity than anything else, no one was really interested in diving too deeply into what he'd done, they were simply entertained that it was still possible. They'd dressed him in a metallic gray suit today, someone probably found the robotic association amusing.
The anchor had been perfectly friendly. Asked him what he did in his time off, and if he had a wife or girlfriend—which Alan took as a sign he hadn't actually read the transcripts of his interviews. No one had seemed particularly interested in his final conversation. He'd made an inquiry as to what had happened to the final bot he'd conversed with, only to be told that the specifics relating to each AI in the tournament were trade secrets, and they were constantly working on refactoring and enhancing their AIs regardless. The woman on the phone promised to get back to him when she knew more. Maybe she'd even call him back. If she really was a person and not another computer.
The cameraman raised one hand to begin counting down from five. He should probably pay attention for this one. The mindshare of this broadcast would probably exceed everything he was going to do in the rest of his life put together.
"Joining us live now, the man who beat the bots. Mr. Alan Turner." The news anchor grinned at him and Alan nodded for the camera politely. "They said it couldn't be done. That they could turn out a bot as good as any human, and no one would ever know the difference. But this man prevailed, in a grueling competition pitting man against machine. So tell me Alan, what's the best way to tell a computer from a human?"
A teleprompter in his field of view flashed a helpful suggestion. “Listen close to see if you hear any gears whirring” If he turned his head far enough he could see the script being fed to the anchor as well. The teleprompter flashed to a different line within a second. “I can’t give away all my secrets, you should stay tuned to the next competition to see what new lines of play open up” Watching the suggestions flit past him, Alan wondered if this was how it felt to be a bot, if all they were doing was picking the best canned line for any situation, layers of primitive intelligence stacked on top of other layers, like a human pulling from their subconscious.
"To answer that question, you have to think about what it means to be human." Alan responded, turning back to the anchor.
The anchor smiled and nodded. "All right. Why don't you tell me about that?"
"It comes down to creativity. Originality. Being able to say or do something that's never been done before. The past and present of the human race is increasingly consumable in digital form, which allows computers to trace back lines of thought on almost any topic and mimic any interaction humans have ever had. Our challenge is to find areas that haven't been explored before, and test the limits of the bots in exploring new ideas, doing what makes us human."
Alan slipped into his own thoughts as the news anchor laughed and continued his chatter. Perhaps he was overthinking things. It was an easy trap to fall into, reading insight and perspective into comments that were nothing more than cleverly stitched together responses, letting your own brain create the illusion of personhood where none existed, like seeing a face on the moon. Kasparov had once been fooled by what was essentially a random guess into overestimating his opponent when playing chess. Maybe the last bot was just programmed to try and say anything if it had been found out to confuse its opponent. The problem was, his gut told him differently. And looking at recent events, it seemed that his gut could be trusted.
But if he really believed that, nothing else in the world made sense anymore. This is how people like Kurt went crazy. You have to believe in an inviolable line between man and machine, something his own consistent performance against the bots should be a testament to above all else. The moment you stop believing that every human carries the divine spark and that every bot is functionally equivalent to a toaster oven, the world no longer makes sense. If machines and men really are on a level playing field, it's no longer as far-fetched to see a machine hiding on every street corner, or feeding suggestions to humans who outsource their judgment and personality to an AI pulling their strings like a puppet.
And if the AIs really were sentient, there were political and legal implications. Should an intelligent bot have the same rights as a human? Should they be allowed to vote? What if there were already billions of them, should they get to run the country? And then there were the ethical paradoxes. Is it wrong to switch a bot off? If there's truly no ghost in the machine, is shutting it down a morally neutral act if you could always bring it back to life in some future era? If the bot enjoys an uninterrupted experience of reality whenever it’s brought back online, does death have any meaning for it?
Alan wasn't sure how long had passed since he'd last spoken, it seemed that the news anchor had chosen to round out the segment with trivia related to the Turing Olympics. It wasn't even very insightful trivia. This kind of thing is why no one had asked a data dump question in years.
"Can I ask you a question?" Alan decided to speak up.
The news anchor stopped mid-sentence, and smiled back at him. "Sure, go ahead."
"It's the kind of question we ask humans and bots to tell the two apart. How would the world be different if news anchors had to pay people to listen to them, rather than being paid to talk?"
Alan waited to see the anchor's teleprompter flash a sentence and quickly started talking before the anchor could speak. "And yes, I realize you'd be broke if that happened." Another sentence flashed on the teleprompter and Alan ran with the lead. "Maybe you would use fewer words, start evolving an economy of expression. But I'm hoping you can suggest something more interesting." The anchor was beginning to look uncomfortable. He tried to start talking right as the next sentence appeared, but Alan was quicker. "And no, I'm not saying you have to pay me for this, so be as chatty as you want."
Whoever, or whatever, was feeding words to the teleprompter appeared to have given up. Alan watched the anchor look back and forth between the teleprompter and him with exasperation. This is usually where he made his decision. Put them on the ropes, demand originality, then see how they respond. Human or AI. Intelligent or unintelligent. Pass or fail. Come on, this shouldn’t be hard, Alan thought to himself. It’s basically just asking what it would be like if you were on the other side of the hustle, paying for ad space. Just show me you’re more of a person than the bots we might be keeping as slaves.
The news anchor flashed a fake plastic smile, and turned his camera ready features away from Alan. "Ha! I certainly wouldn't want to be one of the machines you were interrogating, I can see why you did so well. Let's talk about something else."
Alan slumped back in his chair. Well. There you have it. The anchor continued to prattle on, as if none of his words had made any impact. God damnit.
Each one of your stories reminds me of slightly different things, even if they're on related themes. This one reminded me of Zero HP Lovecraft's "God-Shaped Hole".
It seemed inevitable this was heading towards a human-bot mismatch, but it didn't end up as classical conventions would go - props for that. Plus it's left ambiguous What It All Means: was that really a bot? Or are some genuine humans actually so programmatically rote in some contexts that they themselves can't pass the TT? (Insert modern NPC meme.) Which of course has Implications, since I think we all know the former very much exists IRL right now, no Singularity needed. Any autist can also relate: sometimes even from the inside, "being human" can feel an awful lot like merely predicting the next expected token. It's more complicated with many-layered interactions like body language and social graces, but mere text...it's so easy to escape the Uncanny Valley with text alone. And dialogue, to a certain extent (current voice deepfakes are remarkably accurate, even if the video parts remain jarringly fake).
Nevermind the in-universe complication that now every single "human" coulda actually been a bot. There's nothing in the parameters requiring them not to lie - in fact that's the whole point of the game! And lying convincingly is one of the more important tells that someone is a human...it's actually very hard to retain 100% intellectual consistency, and calling out such "blunders" in real human interactions is a quick way to ostracize oneself as a social loser who Just Doesn't Get It. So it'd make sense to keep around a real human for training purposes, possibly even set them up in a Truman Show-type situation.
Also, it's interesting that Alan apparently has both really excellent eyesight and the ability to quickly and reliably read text backwards. Unless I misunderstood something about teleprompters and camera angles. (Or maybe he's just guessing at the prompts...which is possibly worse, since they'd be uncannily good guesses.)
This was an interesting story, Mark. The premise pulled me in and made me root for the human. I saw scary hints of what a bot-run future could look like.