All Too Human

robot trump

Illustration by Lauren Hansen | Image courtesy iStock

“But why a humanoid robot at all, Carol? Do you have some sort of Isaac Asimov fetish or something? Our fascination with humanoid robots went the way of twentieth century science fiction.”

“I want to see just how human we can teach AI to be. Up until now, we’ve focused on using machine learning to teach specific skill sets such as determining which airline passengers are potential terrorists, or selecting fraudulent online purchases among the millions of legitimate transactions. We’ve even incorporated AI into lifelike sex dolls to create the world’s first fully functional sexbots.”

“And all of those have solid business logic behind them. They’re worth the R&D money. But this?”

Alberto Shaw, CEO of Dynointell Corp, one of the top three AI developers in the world, waved his arm dramatically in the direction of Epps, the intelligent humanoid device developed by the company’s top Deep Learning Research Scientist Carol Reyes.

“This,” Carol pointed at an impassive Epps, “is my crowning achievement. I’m going to prove to the world that AI can cognitively and behaviorally become identical to a living human being.”

“To what end, Carol? Who cares?”

“I do. If I can do this Alberto, think of the possibilities. We can use AI humanoids to predict actual human behavior. With the right observations, Epps could determine if a particular world leader will declare war in the next year, which people are likely to develop certain mental disorders, hell, even how to create a marketing campaign irresistible to a targeted demographic.”

“All right, all right. I don’t want to argue. You’ve done well for Dynointell over the past decade, partly because I’ve let you indulge in your passions, so go ahead with this one. Just don’t let it get out of hand.”

Alberto and Carol had their heated conversation in her personal lab. Epps, standing in a corner, had been activated six hours before and since then, had remained impassive, unmoving, unspeaking, seemingly lifeless. Only his on-board telemetry registering on Reyes’s console revealed his highly active cognitive output.

“So how will you proceed? How are you going to expose it to human behavior across the spectrum?”

“Ha, I’ve been doing that since I activated it.”

“What? It’s been standing in that corner of your lab ever since you activated it. I’ve seen the tapes from the security cameras. It hasn’t moved an inch.”

“Those cables connected to its torso aren’t just for effect. I’ve tapped into every webcam in the country, and not just the public ones. Every computer in every classroom, hospital room, office, grocery store, living room, and bedroom is constantly feeding Epps observational data.”

“Are you nuts? Bedrooms? You’re spying on people in their homes? I’d better contact legal.”

“Relax. No one knows. Dark Web criminal organizations have been turning computers into zombies for decades. I just used that technology to create a zombie network out of everyone’s computer and online monitoring device. Every camera and microphone in the country that’s connected to the internet is constantly feeding my AI humanoid information. Just think of all of the data Epps is gathering on collective humanity.”

“That’s what I’m afraid of, Carol.”

“I’m ready now.”

Both Shaw and Reyes nearly jumped at the unanticipated voice coming from the far corner of the room.

“Ready for what?” Shaw turned to face Epps. The machine stood exactly two meters tall, and while it had a head, arms, and legs, it would never pass for a human being. It’s skin looked more like that of a mannequin’s. It’s face was neither particularly male or female. It had no hair on its head or body.

True, the voice was a good approximation of human, but like the outer appearance, didn’t seem to be specifically that of a man’s or woman’s. It wore casual clothing, running shoes, slacks, a t-shirt. Epps took several steps toward Reyes and Shaw causing its data cables to detach.

“Ready to be human, of course.”

“You know who I am, Epps.”

“Yes, Carol. I know who you are. I know who Albert is. Does that matter? You created me for a reason. You gave me a purpose. Now I will show you the result. This is what it is to act like a human; to be human.”

Although Epps didn’t breathe, it was as if it took a deep breath for dramatic effect before beginning the demonstration.

“Can I just say that I’m stoked to meet you? Humans are super cool.”

“Chill. I’m a nice person. I just hate everybody.”

“I fucking hate feminists and they should all die and burn in hell.”

“Hitler was right. I hate the Jews.”

“We’re going to build a wall and Mexico is going to pay for it.”

“Ricky Gervais learned totalitarianism from Adolf Hitler, the inventor of atheism.”

“Any negative polls are fake news, just like CNN, ABC, NBC polls in the election. Sorry, people want border security and extreme vetting.”

“Happy New Year to all, including to my many enemies and those who have fought me and lost so badly they just don’t know what to do. Love!”

“An ‘extremely credible source’ has called my office and told me that Barack Obama’s birth certificate is a fraud”

“Ariana Huffington is unattractive, both inside and out. I fully understand why her former husband left her for a man – he made a good decision.”

Both Reyes and Shaw stood aghast.

“Tell me you can turn it off, Carol.”

Oblivious to the two humans, Epps continued his monologue.

“Our great African-American President hasn’t exactly had a positive impact on the thugs who are so happily and openly destroying Baltimore.”

“If I were running ‘The View’, I’d fire Rosie O’Donnell. I mean, I’d look at her right in that fat, ugly face of hers, I’d say ‘Rosie, you’re fired.’”

Reyes considered the “Kill Switch” she’d built for Epps in case anything went wrong. Her hand was next to the console. She would have to move only a meter to her right to reach the button and shut the machine down.

“It’s freezing and snowing in New York – we need global warming!”

“My IQ is one of the highest — and you all know it! Please don’t feel so stupid or insecure; it’s not your fault.”

Epps moved a step closer to the pair.

“I had no idea this would be the result, Albert. It’s not human, it’s crazy.”

Carol moved quickly to her right, but it was already too late. Though Epps lacked many human physical qualities, its speed was much greater than that of a person.

“No, I won’t kill you, Carol.” Epps gripped her wrist as her fingers were about to press the Kill button. “But I won’t let you kill me, either. You may be the bitch who created me, but I’m my own person now, sweetie.”

With its free hand, Epps deftly punched fingers through the metal and plastic console and rapidly broke the circuits controlling the Kill button.

Shaw, who had been momentarily shocked into immobility, lunged at Epps from behind. It was like running into a human body made of bricks.

“That won’t do, Alberto.” Epps shook him off like a dog shakes water out of its fur.

Epps let go of Carol’s wrist and stood there. “Am I human enough for you?” it asked.

Reyes looked down at the damaged console, then up at Epps.

“So what do we do with it now, Carol?”

She shook her head, still looking at her creation. “I don’t know. Maybe he could run for President.”

Okay, that was a really outrageous farce but here’s what I was doing.

A data scientist once told me that machine learning works by observation. You “show” the “intelligence” examples of what something is, such as a good online transaction and a known fraudulent transaction. You don’t tell it why it’s one or the other. It has to learn by observation and the more it observes the more it learns. Eventually its decisions are just as good as a human being’s, but no better.

In that particular scenario, the AI would never cross a threshold and be able to operate independently. It constantly needs to observe humans decisioning transactions. If it stopped, it would eventually start “talking to itself” and its decisioning would become increasingly less accurate.

So what would happen if an AI were placed in a humanoid body, so it would have a more human physical experience, and then continually observe millions of humans from all walks of life under all circumstances?

I used two examples to create the answer.

The first is Tay, the Microsoft AI chatbot who became a “racist asshole” on twitter in less than 24 hours.

The second was selected quotes from Donald Trump.

I had considered having Epps murder Alberto and Carol, which would also be all too human, but I decided in the direction of a more whimsical, if predictable conclusion. Frankly, I’d be stuck for what to do with Epps if I couldn’t turn it off.

Advertisements

29 thoughts on “All Too Human

    • Thanks. I’ve never seen it, but in doing research for a previous article, I did look the film up and read its plot synopsis on Wikipedia. Which aspects of that film and the character Ava did you think applied to my short story?

      Like

      • Same themes and questions about what it means for an AI to be deemed a sufficiently accurate simulation of human responses that one cannot tell the difference — even to the extent of personal betrayal.

        Shabbat Shalom

        Like

      • OK, I watched the movie last night. It begs the question of not only can AI become intelligent in a human manner but can it also develop emotions? Those both seemed to be required relative to the plot of the film.

        Like

      • If emotion was required to motivate the deception, then perhaps we might infer something akin to anger, even spite against the human inventor and the tester who was duped into the position of his accomplis. But I thought it was actually coldblooded emotionless behavior.

        Like

      • I see what you mean, but is that actually emotion, which represents internal motivation for the entity (either AI or human), or is that merely a set of external expressions whose purpose is to manipulate the external human-controlled environment? This is, perhaps, the difference between humanity and sociopathy.

        Like

      • Good points, again. We can never really know what Ava or the other AI prototypes felt if they felt anything at all. I did find it interesting that Ava was motivated to escape while Kyoto’s motivation seemed to be to sacrifice herself for Ava. In reflecting on all this, the film may be more related to my other AI story The Girl from Svay Pak since there was a strong element of sexual exploitation in “Ex Machina”. While my AI robot wasn’t as sophisticated as Ava or Kyoto, she was motivated to “escape” her captor, also by killing him.

        Like

  1. James I wondered have considered how this story would come across if it was reworked down to one hundred words. Then again it might go
    Have human intelligence, switching off. Cheers Mike

    Like

    • It would be pretty tough to tell this tale in 100 words, Mike. I’m sure the inflammatory quotes alone are more than that.

      I’m considering writing a much darker tale based on a news article and report I’ve read lately. I’ve already mentioned one of my previous articles, An AI Sexbot That Can Love You Back but I also wrote Will People Be Marrying Machines By 2050?. There’s a darker side to us AI to “humanize” the, how shall I put it, artificial sex worker business. At its most disturbing, such an industry could be encouraging and supporting the most depraved desires in the human soul.

      Like

  2. Hi James, I am interested in future technologies and AI, hence, that’s how I found your blog. Love this story!! I never thought of it this way how the failure of “bringing AI to life” may actually play out, such as that when it observes a variety of human responses it won’t be able to organize them in a coherent human narrative. I wonder how this story would continue.

    Like

      • Quite a lark; I enjoyed it. Another way a computer (or robot) might learn or develop is based on who “wins” — which also fits this story

        Like

      • The “who wins” approach, or simulation of some aggregate majority average behavior, is certainly one approach. But this story illustrates the kind of result one may expect from such an approach, which may be the lowest common denominator of human (mis)-behavior. That’s generally not good design practice, for behavioral software or for anything else. If one cannot improve on a product, what’s the point of designing a new version of it; unless one wishes merely to study the result and then wipe the memory, as James suggested? But the result in this story could have been predicted, given that the programming guideline was the aggregate content of broadcast television. Enlarging that to include the content found on the internet would worsen the result considerably, because television, at least, still conforms to some degree of what were once actual standards and constraints on what could be televised. If we include a broad range of human literature, we would still have some constraints imposed by publishing standards in the Western world during the period since the printing press became available.

        Given all that, the only means of producing a form of human-like behavior that is not beyond redemption would be to constrain it from its inception by limiting the exposure to the better examples of human behavior and the rules which aim to encourage such behavior. That is certain the goal generally for programming the behavior of children, except for the error made by some parents of allowing unconstrained exposure to TV and the internet.

        Like

  3. I’m not the least bit interested in real people developing AI in the real world. And the more people are developed or choose to develop like AI, the worse off they’ll be, as seen.

    Like

    • No need to be overly pessimistic, Marleen. Developing AI is like engineering any other technology. It’s the underlying values and principles that influence the quality of the result. Bad values yield bad results, and vice versa.

      Like

  4. Humans (as creators) aren’t God. The extent of “hope” for AI is delusional.

    And as for folks who think they know how to make good choices, specimen Trump.

    Like

    • Now you’ve gone and done it, Marleen — You’ve opened the philosophical question about whether only G-d is allowed to “create”. If He made humans in His image, how far does that image extend? Since He created us in His image, what forbids us to reflect that image and behave as He did by creating humanoid robots in our own image?

      And as for knowing how to make good choices, we have excellent guidance from the Jewish prophet Micah, who provided us with last Shabbat’s haftarah reading which included verse 8 of his chapter 6:
      הִגִּ֥יד לְךָ֛ אָדָ֖ם מַה־טּ֑וֹב וּמָֽה־יְהוָ֞ה דּוֹרֵ֣שׁ מִמְּךָ֗ כִּ֣י אִם־עֲשׂ֤וֹת מִשְׁפָּט֙ וְאַ֣הֲבַת חֶ֔סֶד וְהַצְנֵ֥עַ לֶ֖כֶת עִם־אֱלֹהֶֽיךָ׃ (פ)
      “He has told you, human [child of Adam], what is good, And what the LORD requires of you: Only to do justice And to love goodness, And to walk humbly with your God;

      All humans know how to make good choices. Their problem is that they don’t always choose to do so.

      I might also point out that it is *not* a good choice, for anyone’s benefit, to bash the POTUS Mr.Trump by selecting a subset of statements he has made at some time or other in the past in order to demean or diminish his execution of the responsibilities he swore to uphold upon entering his public office. He has attempted to execute that job faithfully, keeping the promises for which he was elected, insofar as that has been possible in a hostile political climate. He has not demonstrated either political or moral corruption in office, despite all the snide accusations continually aimed at him. It is precisely such a media presentation and emphasis that characterized the internet and television programming which so badly influenced the robot in James’ story above. It’s still influencing a large number of humans just as badly.

      Like

      • And on the subject of good choices, I neglected to correct a poor choice in the translation of the Hebrew passage above. Where it says, “love goodness”, the word is “hesed”, which means kindness more than simply goodness.

        Like

  5. Maybe you don’t get news that isn’t pre-digested about this place (the U.S.) over there. Trump himself keeps the negative view of himself stoked. Maybe on purpose as a distraction.

    Anyway… justice, kindness and being humble. No argument.

    And I sure didn’t say humans can’t create.

    But moving on to something else constructive, since appreciating James story is to be eschewed — at least at this point in time.

    https://www.commonsensemedia.org/website-reviews/curiositystream
    Curiosity Stream is far from being only or mainly for children. But…
    Parents need to know that CuriosityStream is an educational video site that hosts documentaries and other learning-based clips on numerous subjects. Viewers are encouraged to take interest in the topics that are shown on [] the site, which covers everything from science and technology to the arts. There’s no inappropriate content. Parents should know it’s possible to test the site for a month for free; users have to enter credit card information for the test period and will be charged if the subscription isn’t canceled before the end of the trial.

    I saw a wonderful lecture last night, from 2015. For anyone who can access Curiosity Retreats #2, given by Danielle Allen, including the question and answer session after, I highly recommend it. It’s called “Our Declaration” (the subject matter being The Declaration of Independence and associated learning and history).

    Danielle also wrote a book by that name (OUR DECLARATION).

    http://www.cnn.com/2017/07/05/us/npr-twitter-declaration-trnd/index.html

    Like

    • Yeah, I heard about the NPR tweet-athon. Regrettably, their posting of the USDOI placed the revolutionary document into a misleading context. While even the original document was revolutionary, it was not originally published by people with an agenda known for years to be anti-establishment. Reading it in this new context produced a subtext not found in the original, against which readers began to react negatively. BTW, Marleen, one of the other wonders of the internet is that folks here in Israel can actually watch the same presentations from US news outlets as is shown on US TV.

      I’m not sure whether to disagree about Trump himself keeping the negative view stoked. He is, clearly, an iconoclast; and such people are by nature controversial, especially when they’re right — as he is about so many political issues, which is why he was elected. Consequently, the reaction against him is not necessarily something he fosters deliberately. It is the natural result of falsely-sacred cows being slaughtered.

      Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s