scotto.org
Stories Scripts Videos Music Non-Fiction Books Blog




The Future's So Bright

The 14/48 Projects partnered with Infinity Box Theatre Project to produce 14/48: Centrifuge 2, an event which paired five playwrights and five science writers to create an evening of 14/48 style theatre. This year I was paired with Elle O'Brien, who brought an absolutely fascinating topic to the table that was irresistible to work with, as you'll hopefully see in the resulting play below.

The format of the night was that each science writer presented a talk of up to five minutes, which was followed by each companion play. With Elle's permission, the text of her talk is presented here first to provide the context, then the script of the play follows. My random draw was to write a play for two men and one woman; the randomly drawn theme of the night was "Continuity"; the show was presented May 5-6, 2017.

Mood Ring 2.0
A talk by Elle O'Brien

I’m Elle O’Brien. I’m a science writer and researcher at the University of Washington, where I study speech and hearing. Recently, I encountered an article in one of my favorite journals, IEEE, that looked a lot like what might happen if electrical engineers set out to improve upon the mood ring.

If you’ve been around the last few decades, you’ve probably tried on a mood ring, and you know that it’s a fad product that claims to reveal your emotions by its changing color. Of course, the ring doesn’t know a thing about anger or disappointment or contentment or any other feeling. It’s not magic, it’s a thermochromic liquid crystal. It’s responding to fluctuations in skin temperature, which you would have learned if you squeezed your ring and watched the readout change from “romantic” to “unsettled”.

In the study I read, a group of psychologists and engineers at the University of South Carolina tried to make a proof of concept that you could keep track of someone’s emotional state throughout the day using a couple of lightweight, wearable sensors. They went a little further than just tracking moods, and tried to predict the likelihood of interpersonal conflicts based on the readouts of the sensors. There have been a lot of efforts before to train computers with techniques from machine learning to recognize emotion, whether from speech or biological measures or something else. It’s not uncommon to see studies about that in the world of machine learning. What was notable about this particular study was that it took place in an uncontrolled environment—participants went about daily living, moving about their homes and their city, for a whole day. And, that by synchronizing the readouts from multiple people, the experimenters attempted to predict not just foul moods but actual events, like arguments.

Quite a few measures were analyzed. Electrodermal activity sensors kept track of participant’s sweating, and heart monitors tracked changes in pulse. You’ll note that neither of these measures are directly related to brain states at all-they’re a proxy for what happens out in the peripheral nervous system when someone gets worked up. It might be hard, on the basis of sweat and heart beats alone, to train a computer to distinguish rage from anxiety from erotic excitement.

The experimenters knew this, and tried to capitalize on one of the most informative channels for conveying emotion—human speech. Study participants were fitted with small microphones, which logged their conversations. Then the acoustic data was processed with software that track changes in voice pitch and loudness, two features that we often vary to convey something about our inner states. Of course, there are only so many ways you can modulate your pitch and intensity, and many more emotions than that. Long before machine learning came onto the scene, speech scientists have known that there isn’t a clean and simple mapping from voice frequency to feeling. Of course, there’s more information available: word choice tells us a lot about someone’s mood (or at least, if they’re feeling positively or negatively about the conversation topic), and non-linguistic cues like “uh-huh”, “hmm”, and “ah!” are also pretty useful. These features were also fed into the computational analysis in the study.

So did it work? The experimenters found that their machine learning analysis predicted episodes of conflict reasonably well, although the strongest predictor was not a biological or acoustic measure but a self-report of feeling angry. Their program had too many false-alarms and missed episodes to be ready to send into the real world. Still, the authors considered their proof of concept to be a successful demonstration that wearable sensors could predict brewing upsets some of the time—and not in a sterilized lab environment, but in the noisy, difficult real world. There are surely limitations to what information can be derived from the types of data collected, but it’s reasonable to think performance could even improve by training their machine learning algorithm on a bigger data set.

Of course, even if this technology were better at detecting our inner feelings than we actual humans, there would be some major considerations to address before turning on our new fleet of machine therapists. How would we protect the massive amounts of extremely personal data being collected? What sort of security would be needed for that? And would the technology really be unobtrusive and accurate enough to improve our lives? Will we even like being emotionally monitored?

For one answer to that question, I’d like to introduce our play: The Future’s So Bright.

The Future's So Bright
A play by Scotto Moore

Christabelle and Roger sit next to each other in a car.

CHRISTABELLE: We were in the car home from work when the level one alarm went off.

An intense beeping sound. Christabelle checks her wristband, shuts off the alarm.

ROGER: What’s that?

CHRISTABELLE: I’m supposed to meditate.

ROGER: Why?

CHRISTABELLE: I had a bad day at work.

ROGER: Do you want to meditate now?

CHRISTABELLE: I can meditate when we get home.

ROGER: Are you sure?

CHRISTABELLE: Roger-

ROGER: See, you do seem anxious.

CHRISTABELLE: You’re making me anxious.

ROGER: Which is probably what it’s predicting, so why don’t you just take a moment right now, I’ll be quiet, and you can clear your mind before we get home.

The two of them transition to a living room as Christabelle continues speaking.

CHRISTABELLE: But when we got home, it wasn’t much better. We were watching television, and I don’t know, something might have triggered me on the show we were watching, or maybe I was just, you know, holding onto something from work, but suddenly my monitor went off again.

An intense beeping sound from Christabelle’s wrist.

ROGER: Again? Is something wrong?

CHRISTABELLE: Nothing’s wrong.

ROGER: We could watch something else.

CHRISTABELLE: Maybe it’s glitching.

ROGER: It doesn’t glitch, they’ve got redundant AI clusters for that very reason. You know you can tell me if something’s bothering you.

CHRISTABELLE: Nothing’s bothering me.

A different intense beep - this one’s coming from Roger’s wristband.

ROGER: See, now I’m getting upset too.

CHRISTABELLE: Yeah, I didn’t need an AI cluster to notice that.

ROGER: Well, I’m going to take five minutes to calm myself down and then maybe we can have a civilized conversation about whatever it is that keeps triggering your monitor.

Roger closes his eyes. We transition to a flashback - an office.

CHRISTABELLE: What happened was, we wanted a really fancy wedding. But we were poor at the time, and Roger’s parents would only pay for such “extravagance” if we committed to being monitored by this technology company that Roger’s dad founded. The system was called “Preventative marriage monitoring,” designed to reduce the likelihood of divorce.

ANGUS: My name is Angus. I’m your assigned technician slash therapist. Here’s how this will work. You’ll each wear galvanic monitors, kind of like Fitbits, but instead of monitoring the amount of exercise you get in a day, we’ll be monitoring your electrodermal and electrocardiographic activity - basically we’ll be able to detect via skin temperature, sweating, heart rate and so on when you’re experiencing heightened stress.

ROGER: But how can you tell the stress is related to our marriage?

ANGUS: Well, we can tell when your monitors are in the same room, and the monitors have microphones in them, so we’ll be monitoring your conversations as well.

CHRISTABELLE: You’re going to be eavesdropping on us?

ANGUS: Well, not me personally, but our computers will be listening. We’ve got networks of artificial intelligence that will scan your conversations, listening to your intonation, pitch range, volume, as well as picking out key phrases - for instance, “you” statements are more likely to be accusations than “I” statements in a conversation, arguments tend to use dramatic language like “always” or “never,” that sort of thing. Don’t worry - actual people only start listening when there’s an emergency in your marriage.

Angus exits. Transition back to the living room.

CHRISTABELLE: So first the system tries to get you to relax, telling you to take a deep breath, or meditate for fifteen minutes, or listen to some new age music, whatever works for you. Then level two kicks in - if you’re still upset with your partner, you’re supposed to drop what you’re doing, and sit together, and just stare into each other’s eyes or whatever until you both calm down and just sort of see each other, you know. It’s supposed to stop you from stabbing each other I guess. But then if that doesn’t work, for level three, they send out an on-call technician slash therapist.

Doorbell rings. Roger goes and opens the door - it’s Angus.

ROGER: Angus! What a surprise. We’re not even fighting!

ANGUS: That’s why I’m here. The system is predicting that you’re about to have a major fight and they, uh… well, I’m on call tonight, and I’m, uh, familiar with your account, so… here I am!

ROGER: Well, come on in, let’s get to the bottom of this. You know, Christabelle has been preoccupied tonight, but so far she hasn’t said why. Maybe you can talk to her for me?

CHRISTABELLE: Behind the scenes, artificial intelligence systems are learning all about your behavior from all of the signals you’re sending back to the mothership. Machine learning algorithms are getting better and better at figuring out how you work. And they’re comparing your data to thousands of other couples because really, aren’t we all the same when you get right down to it?

ROGER: Sweetie, look who’s here, it’s Angus.

CHRISTABELLE: Yeah, I know it’s Angus.

ANGUS: Hi, Christabelle.

ROGER: If you won’t talk to me, maybe you’d like to tell Angus what’s bothering you?

CHRISTABELLE: Maybe Angus can tell you himself.

ROGER: I’m sure I don’t know what you mean.

ANGUS overlapping: Uh, I don’t think that’s such a good idea.

CHRISTABELLE: Fine, I’ll say it. Angus and I have been fucking each other since right after we met.

After a beat, both Christabelle’s and Roger’s wristbands start beeping very loudly.

ANGUS: Uh, let me just shut those off for you…

Angus operates a small tablet and the beeping stops.

ROGER: How is that possible? If Christabelle was so unhappy with our marriage that she was cheating on me, the system should have predicted that months ago, right, Angus?

ANGUS: Well, uh… I’ve been editing your marriage’s continuity data a little bit.

ROGER: What? How is that even possible without triggering deeper alarms in the system?

ANGUS: It’s very very difficult, I can assure you, but Christabelle of course is very very worth the effort, and the risk, which I’m surprised you yourself haven’t realized yet…

ROGER: When my father learns that the system can be compromised by some random technician-

ANGUS: Random? None of this is random! I’m the technician your father specifically assigned to your marriage, Roger! And my editing of your marriage’s continuity data is the only thing that kept your marriage intact this long in the first place!

CHRISTABELLE: Which begs the question, what exactly have the machines learned about us? How to prevent conflict? How to generate conflict? How to push us around for no apparent reason except their own amusement? I’m telling you, this is how the robot apocalypse starts.

ANGUS: I should have known I couldn’t hide this forever. The system figured out I was interfering with its algorithms. I don’t even know how it knew - it locked me out as soon as it figured out I was hacking it. But Christabelle - you didn’t have to say anything to Roger! You could have made up some excuse for why you were upset! The system’s not a lie detector!

CHRISTABELLE: Maybe I’m sick of lying about my feelings in the first place, Angus.

ROGER: If word got out that “preventative marriage monitoring” couldn’t save the marriage of the founder’s son, the whole company would die a quick death. And I’d have to pay father back for the cost of the wedding.

CHRISTABELLE: Sounds like quite a dilemma, Roger.

ROGER: Let me ask you something, Angus. You’re sending reports about our marriage to my father, aren’t you?

ANGUS: Of course.

ROGER: Edited reports, is that correct?

ANGUS: Yes, absolutely.

ROGER: Then let’s make a deal. I want you to do more than just edit our marriage’s continuity. I want you to substitute someone else’s entirely. He rips off the wristband. I don’t want to hear that thing beep at me ever again, do you understand me?

ANGUS: I understand completely!

Roger goes to Christabelle.

ROGER: Christabelle - I don’t know where things went wrong between us.

CHRISTABELLE: Probably when you agreed to a total surveillance state for our marriage.

ROGER: I want to make it up to you. Maybe we could start over, as though our marriage had never been monitored.

CHRISTABELLE: You can’t prevent Angus from being a part of my life now.

ROGER: Maybe Angus could learn to share.

ANGUS: Oh I don’t know how I’d feel about-

ROGER: If he wants to keep his job.

ANGUS: That sounds like a fascinating option.

CHRISTABELLE: Well I’m glad that money has played such an important role in solving our problems here.

She removes her own wristband.

CHRISTABELLE: A few years later, Roger’s dad was bankrupt. Turns out we weren’t the only ones who had a shall we say negative experience with preventative marriage monitoring. And I learned that I didn’t feel like choosing between Roger and Angus. The three of us live a gloriously unmonitored life now, and what goes on with us behind closed doors is none of your business.



Copyright Scotto.org until 2087