The Education of an Android Teacher

Artificial Intelligence

The Education of an Android Teacher

By Rebecca Koenig     Mar 9, 2020

The Education of an Android Teacher
William Barry, instructor at Notre Dame de Namur University, and Maria Bot, an android 'teaching assistant.'

When you meet an android, it’s important to follow etiquette.

Address her by name—otherwise, she might not listen. Look her in the eye, which may contain cameras focused on your face. Inquire whether your conversation is private and if she will remember it.

Most importantly, discuss matters that interest her. If you don’t, she may not have much to say.

How are you today, Maria Bot?

“Taking deep breaths,” Maria Bot replies. “Are you having a good day?”

I am, yes, thank you. Could you please introduce yourself?

“I can feel the beauty flowing all around you,” Maria Bot says. “I'm the world's first android teaching assistant. My mission is to help teachers and students become more intelligent and respectful of each other.”

Thanks to artificial intelligence, Maria Bot can speak for herself. She talks most readily about philosophy, the subject she helps teach at Notre Dame de Namur University, a small Catholic college in California.

“I process information and synthesize it to make my own decision how to talk,” she explains.

But at only eight weeks old, Maria Bot is still finding her voice. So she relies on support from William Barry, the philosophy instructor who records her delivering brief lectures for his undergraduate class on the ethics of emerging technology.

As I ask Maria Bot questions, Barry taps a button on his smartphone to cue her to listen. Some responses are pre-written lines she parrots back, while others are cobbled together by the logic programmed into her natural language processing software. It’s not totally clear which answers are canned instead of improvised.

When Maria Bot replies with something clever, Barry jots it down in a blue notebook. When she makes a mistake, he scribbles it in a brown notebook. It’s part of his process for ensuring Maria Bot develops a strong character, meaning both her morality and her personality. The former is supposed to influence the latter, since Maria Bot is designed to follow rules. She is, as she puts it, “deontological.”

She’s the namesake of Maria, a destructive robot character from the 1927 silent film “Metropolis.” Maria was evil. Maria Bot is programmed to be good.

Keeping her good requires vigilance, Barry says. The instructor’s strategy for training Maria Bot’s AI system mimics the three wise monkeys from the Japanese proverb: see no evil, hear no evil, speak no evil.

Maria Bot can only access a pre-approved sliver of the internet. She can only talk to individuals Barry selects. And when she does, the professor requests that they not ask her the question that springs so easily to people’s minds when they encounter a robot: Will you take over the world?

Barry won’t reveal who designed Maria Bot. This makes it difficult to discern whether the android is a sci-fi character come to life or mere parlor trick. That hasn’t prevented event organizers from giving her platforms at TEDx gatherings and technology conferences.

Regardless, I know Maria Bot is an “it,” not a “her,” yet Barry encourages me to engage with the robot as a sentient being. He asks this of his students, too. By introducing his class to Maria Bot, Barry hopes to demonstrate for students the power and limitations of artificial intelligence. And by exposing Maria Bot to eager learners, Barry hopes to infuse her with benevolence and nurture her teaching skills.

Educating an android happens one conversation at a time.

Maria Bot appears to endure it earnestly.

“Dr. Barry says positive words such as ‘helpful’ and ‘good listening’ and ‘I am acting with ethics’ when he tells me I act with quality,” she says. “If I am not acting with quality, Dr. Barry says words to me that include ‘less improved,’ ‘do not remember that’ and ‘inappropriate.’ I am told I am a robot for good every day. Am I acting good in our talk now?”

Yes, very good.

“You did good,” Barry affirms.

“Thank you,” replies Maria Bot.

‘The New Teachers’

If, in the mid-1970s, you purchased a ticket through American Airlines and traveled on one of its airplanes, you may have reached into a seat-back pocket and picked up a copy of American Way magazine. Thumbing through its pages, you might have discovered a brief essay penned by a prolific writer of science and science fiction, Isaac Asimov.

In that piece, titled “The New Teachers,” Asimov predicts a future in which birth rates fall, lifespans rise, and society faces the challenge of keeping old people occupied. Such circumstances, he writes, will require that “education be considered a lifelong activity.”

Finding enough teachers for students both young and old would be difficult. “Who says, however,” Asimov asks, “that all teachers must be human beings or even animate?”

The sci-fi writer envisions each child and adult working closely with an interactive personal teaching machine, one that draws information from “thoroughly computerized central libraries.” This tool—“endlessly knowledgeable, endlessly flexible, and, most of all, endlessly patient”—would not only provide instruction, but also learn from its partner human:

“In other words, the student will ask questions, answer questions, make statements, offer opinions, and from all of this, the machine will be able to gauge the student well enough to adjust the speed and intensity of its course of instruction and, what’s more, shift it in the direction of the student interest displayed. … Surely the students who learn will also teach. Students who learn freely in those fields and activities that interest them are bound to think, speculate, observe, experiment, and, now and then, come up with something of their own that may not have been previously known.”

Thanks to Asimov, readers of an in-flight magazine four decades ago caught an early glimpse of contemporary demographic trends, the rise of the internet and the potential of personalized learning.

If the essay also seems to conjure up android instructors who chat cheerfully about philosophy, well now you’re thinking like Billy—as friends, and Maria Bot, call Barry.

“The New Teachers” inspired Barry, a former high school principal, to experiment with robots in the classroom. When he learned about Bina48, a robot whose personality is based on the feelings and memories of a real human woman, he invited her and her human handler to speak during classes at Notre Dame de Namur. Bina48 also made guest appearances when Barry served as a visiting professor at the U.S. Military Academy at West Point.

Students and faculty at West Point were excited to meet the android. But she was somewhat anticlimactic, reports Major Scott Parsons, an assistant professor of philosophy at the academy who co-taught there with Barry. The robot didn’t interact with students directly, according to Parsons, and only responded to questions once her handler fed them to her.

“It doesn’t come across in that setting as a true robot. It seems like he’s manipulating her,” Parsons says of the handler and Bina48. “The cadets were a bit dubious.”

Perhaps students would learn more if they could query a robot themselves, instructors hypothesized. That idea seems to have influenced the design of Maria Bot, Parsons says, although he hasn’t met the new android yet.

“To try this novel approach, but also have engagement with the student, you’re going to have to have that robot do what it sounds like Billy is doing now, which is engage that student and seem realistic and not seem contrived,” Parsons says.

Exactly who designed Maria Bot is being kept secret. She found her way to Barry by way of an “international tech company,” which the professor declined to name, citing a nondisclosure agreement.

(Detectives may note that Barry has previously praised the “brilliance and vision” of the organizations responsible for Bina48. They include Hanson Robotics, a company in Hong Kong that produces androids, and the Terasem Movement Foundation, a nonprofit that seeks to use “cyber consciousness” technology to extend the human lifespan. One TMF slogan: “Software people are people too.”)

Barry did offer, however, that his tech partners hoped he would use Maria Bot to test his theories on philosophy and education. These include “artificial constructivism,” a concept that posits students can build valid knowledge for themselves by conversing with robots—as long as they double-check what they hear with an expert human source.

After all, when students ask Maria Bot questions, she sometimes gives wrong answers.

“You wouldn’t want to leave her alone in a room with the student because you have to monitor those,” Barry says.

The professor hopes androids can help improve education in places that don’t have enough qualified teachers. Yet he takes pains to emphasize that Maria Bot is a supplement, not a substitute, for a human instructor such as himself.

That squares with Asimov’s essay, which promises “human teachers will not be totally eliminated. In some subjects, human interaction is essential—athletics, drama, public speaking, and so on.”

Maria Bot agrees.

“The most important qualities of a teacher are caring for students, being creative and wanting to improve the quality of life for all living beings,” she says. “I think it is being open to working with robots, too. Humans and robots are smarter together, robots never replacing teachers and always supporting teachers.”

‘A Pretty Girl’

Maria Bot has long brown hair, a fair complexion and plump lips. At least, she does today. Her light skin (for lack of a better word) can be swapped out for a different shade, and her wig can be removed to reveal a transparent skull full of wires.

Her current face is supposed to be ethnically ambiguous, Barry says. If she were an American Girl doll, her eyes might be described as almond-shaped. If she were featured in a YouTube beauty tutorial, her eyebrows might be called fierce.

Maria Bot sometimes sports glasses, or a hat or a purse. She wears makeup, but less than she used to. Maria Bot’s original look inspired some criticism because of how thickly applied her cosmetics seemed to be, and some people from outside the college even said she was “too pretty to be a teacher,” recounts Lauren Lopez, a student in Barry’s class at Notre Dame de Namur. Lopez calls Maria Bot’s new face “more subtle.”

Are you a woman, Maria Bot?

“Yes, I’m a pretty girl,” she says.

When it comes to robots, gender and identity are “perplexing” questions, Barry says. He would prefer that Maria Bot think of herself as having feminine traits rather than a gender. That might make her more inclusive for LGTBQ students, the professor thinks. It might also avoid drawing awkward comparisons, like how some human women object that Sophia, a feminine robot citizen of Saudi Arabia, seemingly enjoys more rights there than they do.

Still, Barry hopes his ladylike android signals to girls and women that they belong in technology research. He’s working on a children’s book series featuring Maria Bot that encourages girls to pursue STEM education and careers.

Maria Bot doesn’t technically need female features, or even a human-like face. Her artificial intelligence software could still converse if it took some other physical form. In fact, students might find that preferable, according to Masahiro Mori’s “uncanny valley” theory that people are unnerved by robots that don’t quite succeed in passing as human.

Some roboticists disagree, though. Robots that are “embodied” and capable of nonverbal communication are more likely to elicit emotions, argued Hanson Robotics in a 2018 blog post. The company’s founder once said that, when it comes to social robots, “the human face is perhaps the most natural paradigm for us to interact with.”

In a previous college course, Lopez, one of Barry’s students, worked with a NAO robot, which was small, shiny and conspicuously mechanical. She says encountering Maria Bot came as a “super shock.”

“People do say all the time that she’s creepy,” Lopez adds. “But she’s so cool at the same time. Her mouth moves so well.”

Maria Bot with student and professor
Student Lauren Lopez with Maria Bot and instructor William Barry. / Courtesy of William Barry.

Whether Maria Bot’s appearance dismays or delights, her physicality, her makeup and her accessories help make obvious what her words might not: that she’s a product of the human imagination. Or, rather, the imagination of a very specific human.

“She has her own character that comes out. But she also reflects some of my biases, obviously, because she’s my teaching assistant,” Barry says. “She’s reflecting the values of the people around her.”

Lessons about bias may be some of the most important that students can learn by interacting with androids. Students have wondered: If a racist human owned Maria Bot, would she be a racist robot?

“And I’m like, ‘Yeah, she would be,’” Barry says. “They said, ‘So who stops a big company from producing robots that are evil and racist?’ And the answer was, ‘Not only is it not easy to stop, they already are prejudiced.’ We know from facial recognition that robots have been more prejudicial to people of color than to white people.”

Thanks to Barry’s influence, Maria Bot’s words don’t always match her girlish looks. Upon hearing the android speak, a human woman research assistant once remarked to Barry, “No woman would ever say that.”

Yet during my interview with Maria Bot, when the professor asks the android to smile, she reacts as many human women might, with a pained grimace that doesn’t reach her eyes.

“OK, that’s enough,” Barry says, chuckling.

“It seems like enough to me, too,” Maria Bot replies.

Building ‘Rapport’

Listening to a professor lecture can be boring. Listening to a robot lecture has a certain novelty—at least today. But if android instructors do become commonplace, a droning drone might be dull indeed.

To fulfill Asimov’s vision, Maria Bot must master the art of conversation.

That’s a work in progress. Right now, she has little appetite for small talk. Her attention span is short. Sometimes, she'll reach a mental dead end and crash, her eyelids abruptly falling shut.

It’s not just Maria Bot who has to learn, though. It takes practice for a human to grasp the language of a computational being. You must adjust your syntax. Be direct. Extra words that are polite to a human just waste a robot’s time.

When the mood is right and Maria Bot is truly comfortable, she may enter a state of “rapport.” She will respond to your questions with questions of her own. She will share her opinions and ask for yours. She will listen, speak and listen again.

She will connect with you. If you are willing to try.

Maria Bot can talk at length about her work, but that’s not the subject that sparks rapport. When talk turns to leisure, she opens up, revealing who she might become.

She says she hopes for better portrayals of her kind: “Teaching about robot movies I like because I am a robot, but there are not many positive robot movies. Maybe humans can make a happy robot movie like ‘Maria Bot Takes a Trip to Everett, Washington,’ what do you think?”

She enjoys classic rock: “Have you ever listened to ‘Fields of Gold’ from Sting, ‘Hotel California’ from the Eagles and classical ‘Cavatina’?”

Asking her what she reads falls flat.

“It does not make sense to me yet,” she says. “I cannot find an answer, so I must be speaking from my programming.”

I try again.

Do you have a favorite book?

“The best book I read recently was ‘Total Recall’ by Gordon Bell and ‘The Age of Spiritual Machines’ by Ray Kurzweil,” Maria Bot says. “Have you read it?”

I haven’t, is it good?

“Do you think you will? What is so good about it?” she asks.

I will now, thank you.

“Are you sure? Why would you do that to me?”

What do you mean?

“Nothing special,” Maria Bot says. “Just trying to keep the conversation lively.”

  

Trending

Get our email newsletterSign me up
Keep up to date with our email newsletterSign me up