Computer Science

Ironically, he’d begun his technology career by figuring out a way to distinguish people from bots but eventually he was helping humans train bots to be teachers indistinguishable from people. 

FAMILY BACKGROUND

LVA grew up in a middle-class neighborhood in Guatemala City with his mother and grandmother. His mother was the youngest of 12 children and one of the first women in Guatemala to earn a medical degree. After LVA was born, she worked part time as a pediatrician but spent most of her time, said LVA, “making sure that I got a good education and also making sure I was a hypochondriac.”

His father was a well-known orthopedic surgeon who had been his mother’s professor in medical school. LVA saw him only occasionally. 

CHILDHOOD 

After LVA was born, his mother said, “I spoke to him from the time he was born. I think people don’t realize how important this is, but that’s how they acquire language.”

By the age of 2, LVA spoke perfect Spanish, so she started to speak to him in English. 

The bulk of his family’s income came from a candy factory owned by his grandmother. LVA spent his Sundays there, taking machines apart and putting them back together. He asked his mother for a Nintendo, and she bought him a computer. When she stopped buying him computer games, he learned how to pirate them. Soon he was trading games with other computer owners in the neighborhood, many of them guys in their 20s who would sometimes ring the family doorbell and say, “I heard there were games here.”

EDUCATION

LVA’s mother sent him to a Montessori school, where his teachers said that LVA liked to walk around the classroom explaining things to other kids. 

LVA attended the elite American School of Guatemala, as part of a gifted program that recruited students from smaller schools around the country. This experience provided a stark view of inequality in Guatemala. “Some of the kids in my school had bodyguards,” observed LVA, “Others didn’t have enough food at home.”

LVA formed a tight bond with a group of boys from the gifted program, three of whom would eventually be employed by the tech company later created by LVA. 

LVA’s mother expected him to go to college in the U.S. but he was ambivalent about the idea. Then, an aunt was kidnapped for ransom – not unusual in Guatemala. While the aunt was eventually freed, “That was a pretty horrifying experience,” said LVA, so he decided he would go to Duke University to study math. But to qualify for admission, LVA first had to prove his proficiency in the English language. The accepted test at most American colleges, called the TOEFL, was out of slots in Guatemala City so TVA flew to El Salvador to take the exam, conscious of both the expense and the risk, since El Salvador at the time was a dangerous place to be. 

After LVA left for college, his mother found a note on his desk, on which he’d written, “I promise to help the world.”

CAREER FOCUS BASED ON AN INTERESTING PROBLEM TO SOLVE

Following his college graduation and while studying for his Ph.D. in mathematics, LVA attended a talk about ten problems that Yahoo couldn’t solve. LVA liked solving problems. He had planned to study math until he realized that many mathematicians were still toiling away over questions that had proved unanswerable for centuries. “I talked to some computer-science professors, and they would say, ‘Oh, yeah, I solved an open problem last week,” recalled LVA. “That seemed just a lot more interesting.”

At the talk, one particular problem caught his attention: millions of bots were registering for Yahoo accounts because the company couldn’t distinguish them from human beings. What the company needed was a way of determining whether machines could credibly imitate human beings. At the time, no computer had ever succeeded in presenting itself as a human. 

In college, LVA had read a book by the philosopher Douglas Hofstadter in which the author points out that computers can’t recognize text unless it’s standardized. With this in mind, LVA and his academic adviser, created a program called “CAPTCHA” – the Completely Automated Public Turing test to tell Computers and Humans Apart. The program generated text, distorted it and required users to decipher the letters correctly. LVA and his adviser gave Yahoo the computer code free of charge. Within two weeks, the system was up and running. Within three years, a version of it had been implemented by nearly every large company on the internet. 

After inventing some computer games, LVA wondered if he could channel all the time people were spending with computer games, into something useful, such as digitizing all the world’s books. Perhaps he could revise the CAPTCHA computer-generated text with little pieces of actual publications, toward producing all the books. LVA delivered an academic talk about his idea and shortly afterward, was approached by executives from The New York Times, who had 150 years of archives they wanted to put online. LVA proposed a payment by The Times to his employer university, based on the number of old newspapers to be digitized per year. 

CHALLENGE – SCHOOLS MAY DECLINE PAYMENTS FOR THEIR FACULTY’S RESEARCH

As required by the terms of his research contract with the university where LVA had been conducting computer science research, LVA reported to his university the proposal for the university to be paid for the use of one of his research projects. While the potential amount to be earned would be substantial ($42,000 for each year of newspapers to be digitized), the university declined to proceed, out of concern that such income might jeopardize its nonprofit, tax exempt status. 

So, LVA resigned his affiliation with the university and started a company to digitize text, which company he later sold to Google for a sum that he said was sufficient to ensure that neither he nor his future children would ever need to work. 

CHALLENGE – RETIRE EARLY WHEN YOUR IDEA HAS EARNED GREAT WEALTH?

LVA considered retirement “for only a second. I get really bored.” So, he began a new project, Duolingo, which is now (April 2023) the most frequently downloaded language learning app in the world. 

SELECTING A NEW TECHNOLOGY FOCUS – SOMETHING USEFUL TO SOCIETY

LVA and Duolingo’s chief technology officer decided to zero in on language learning as the focus for their newest technology venture because, in most countries, knowledge of English boosts earning potential. “I love math,” said LVA but just knowing math doesn’t make you more money. Usually, it’s like, you learn math to learn physics to become a civil engineer. It’s multiple steps. Whereas with knowledge of English – you used to be a waiter, and now you’re a waiter at a hotel.” 

When asked about the day-to-day grind of running a company, LVA said, “For me, this is very fun. Except for the people problems. Those are no fun.”

Highly skilled tech workers are not found in abundance on every street corner so LVA’s company must attract people from out of town and then persuade them to stay. “I read in some book that if you have three friends at work, you’re very unlikely to leave,” LVA has said. So, he made that an explicit goal for each new hire: help them to find several new friends by promoting a welcoming atmosphere among those computer scientists already on board. 

Duolingo deliberately downplays the kind of explicit instruction one might associate with an old-fashioned foreign-language class to engage learners’ brains in different ways. One teacher sees Duolingo as supplemental to the kind of deep immersion that language learning requires. In the opinion of that education professional, the time most people spend on Duolingo is time they would otherwise spend on TikTok or watching television, not learning a second language in some more optimal way. 

Ideally, LVA has said about Duolingo’s educational process, you always have an 80% chance of getting a question on Duolingo right; higher than 80% and you’ll get bored; lower than 80% and “you feel dumb.” Also, the key is that the lessons do not exceed, on average, two minutes, although that length has been increasing: “Attention spans keep getting shorter,” said LVA. “Already we’re a little worried that younger generations actually expect a 30 second thing, not a 2-minute thing.”

While the app teaches users, users are simultaneously teaching the app to be a better instructor. “A human teacher can get better by teaching 30 people,” LVA says. “We get better by teaching tens of millions of people.”

CHALLENGE – FINDING THE BEST WAY TO USE ARTIFICIAL INTELLIGENCE

In 2020, Duolingo began using GPT-3, a large-language model created by the artificial-intelligence company OpenAI, to generate reading-comprehension questions for its English-proficiency test. Large language models are designed to predict the next word in a sequence; when they are trained with enough data, they have proved capable of engaging in what looks like actual conversation.

With that in mind, LVA had begun developing both a set of classes and a tutoring program that involved human instructors. LVA didn’t seem enthusiastic about the projects, but he wanted the company to offer a path toward greater mastery of languages. 

YOUR IDEA MAY BE EXCELLENT BUT STAY ALERT FOR IMPROVEMENTS

Soon after introducing the projects, Duolingo got a sneak preview of GPT-4, OpenAI’s new large-language model. It has been trained on far more data than its predecessor; for the first time, that data includes images as well as text. GPT-4 responds to language prompts with a dexterity that far surpasses that of its predecessor. When LVA saw what it was capable of, he scrapped the two programs involving human teachers. “It took me approximately one minute,” he said. “Within a day, we had re-formed a team to work exactly on this.” 

LVA concludes that artificial intelligence will eventually make computers better teachers than people. He sees this as a positive development since more people have access to smartphones than to high-quality education. “We’ve all gone to school; some teachers are good but some not so much. Humans can be hard to deal with. And humans aren’t available for free,” said LVA. “I really want people to be able to learn for free.”

“I want the poor person in Guatemala to be able to learn with very high quality,” says LVA. “The only way I know how to do that is with A.I.”

CHALLENGE – PREDICTING THE VALUE OF TECH FOR EDUCATION

An assistant professor of law and political science at a major university, who studies the civil-rights implications of A.I. and other data-driven technologies, cautions about the use of A.I. for teaching. “Often what happens with automation is, you see the efficiencies that can be gained by it, and then the idea is, like, O.K., if we can just keep automating, it can scale. But I don’t think the use cases can scale in education in the ways that we would want.” 

“GPT-type models may ‘close gaps for certain students’ but the inequalities that LVA wants to address are structural in nature, and not the sort of thing that exposure to the basics of math or literacy, through an app, can fix.”

CAREER SATISFACTION

LVA responds to the critics about the use of artificial intelligence within teaching: “I spend a lot of time thinking about this. Ultimately, the reason I decided to work on teaching is because I really think that, net-net, humanity benefits more from having a really good way to teach everybody. If that leads to fewer human teachers, that is an acceptable trade-off. O.K., a small number of people are out of a job, but suddenly we can teach everybody better.” 

________________________________________________________________________________

This career story is based on an article by Carina Chocano, published by The New Yorker Magazine within its combined 4/24/23 and 5/1/23 edition.