The Future of Education: Online, Free, and With AI Teachers?
As K12 and higher education evolves to incorporate immersive technologies like AR and VR in the classroom, educators should also keep their eyes on AI and machine-learning trends to enrich the student experience. Read on to see how the Co-founder and President of Duolingo has found opportunity for AI in learning.
While it might not get a lot of investor attention, education is actually one of America’s largest markets.
The U.S. has 20 million undergraduates enrolled in colleges and universities right now and another 3 million enrolled in graduate programs. Those undergrads paid an average of $17,237 for tuition, room, and board at public institutions in the 2016-17 school year and $44,551 for private institutions. Graduate education varies widely by area of focus, but the average amount paid for tuition alone was $24,812 last year.
Add all of those up, and America’s students are paying more than half a trillion dollars each year for their education! And that doesn’t even include the interest amassed for student loans, the college-branded merchandise, or all the money spent on beer and coffee.
Keeping the costs down
Several companies are trying to find ways to make college more affordable and accessible.
2U is partnering with universities to offer graduate education online. While the tuition is unchanged, online programs don’t require graduate students to move their families across the country just to be on campus. 2U CEO Chip Paucek recently told The Motley Fool that he believes we’re still in the early innings of online education. 2U has expanded its reach in recent years and now offers programs from 34 different universities.
But another way to take costs out of our education system might be to replace the human professors altogether.
Duolingo is a website and smartphone app that helps 300 million people across the world learn new languages. It’s absolutely free. It uses an artificial-intelligence engine to deliver the lessons to students. To put that number in perspective, there are now more people learning languages on Duolingo in the United States than there are people learning languages in the whole U.S. public school system.
That AI engine is worth double-clicking on. Duolingo is able to adjust how it teaches — for each individual student — according to the feedback it receives. If students who learn about plural nouns before adjectives are progressing faster, it can adjust the lesson plans. The goal is to maximize the overall proficiency of all students.
The AI doesn’t require a salary or health insurance. But it is able to personalize lesson plans, account for the fact that students learn at different rates, and teach at a consistent level at a global scale.
This could be a quantum leap for education. The United States is fortunate enough to have a rather well-developed education system, which is structured and can attract talented educators. But in many developing countries, that’s often not the case. AI could level the playing field for many who might not otherwise have access to formal schooling or to good teachers.
Looking one step ahead
But teaching foreign languages is also just the beginning. AI’s grander ambition, which Duolingo will certainly support, is for computers to learn how to better understand how humans speak.
That could unlock a ton of value for the enterprise and for investors. Conversational AI could help consumer-facing companies like Amazon better understand and react to their customers’ needs. Retailers and hospitality companies like Marriott are deploying hospitality robots to help guests place orders or get directions. Synthetic biology is beginning to use AI to design customized proteins for companies like Bayer to produce higher-value chemicals.
In other words, AI is raising the bar for a variety of industries. And the companies that aren’t actively exploring AI could eventually become obsolete.
To get a closer look at the opportunity for AI, Motley Fool Explorer lead advisor Simon Erickson recently spoke with Duolingo founder and CEO Luis von Ahn. Luis is a famed inventor who previously founded CAPTCHA (later acquired by Google), and his work on Duolingo led to him winning this year’s prestigious Lemelson-MIT prize.
In the interview, Luis describes the state of artificial intelligence today, what we should expect from it in the future, and why the next wave of computer-science students is interested in using AI for societal good. (He is also an associate professor at Carnegie Mellon, which is a prominent computer-science school in Pittsburgh.)
A full transcript of the conversation is provided below.
Simon Erickson: Hi, everyone, I’m Simon Erickson, Motley Fool Explorer lead advisor, and I’m joined today by Luis von Ahn. Luis is the co-founder and president of Duolingo. He’s also the winner of this year’s MIT’s Lemelson Prize, which is the largest cash prize for inventors. Luis, thanks very much for joining me here this morning.
Luis von Ahn: Thank you for having me, Simon.
Simon: We’re here at MIT, and of course a lot of cool inventions going on. But one of your biggest accomplishments that we just mentioned was Duolingo, which is helping more than 300 million people across the world learn different languages. Tell me a little bit about this project that you’re working on.
Luis: Yes, so we launched Duolingo about six years ago. It’s an app to learn languages. For people to learn languages. The idea is that you can learn languages for free. And it’s very effective also. When people use it they really are learning. We have used this all over the world like you said, 300 million. It’s the largest way to learn languages in the world with the most popular way to learn languages in the world. There’s a lot of interesting stats. For example, there are more people learning languages on Duolingo in the United States than there are people learning languages in the whole U.S. public school system.
We teach a number of languages. We teach Spanish, English, German, etc. But we also teach some smaller languages. For example, we teach Irish. It turns out there are 94,000 native Irish speakers and we have about a million people learning Irish on Duolingo. So we can actually multiply it by 10, the number of people who are speaking that language when we put it in the platform.
Simon: Sure. Now we said, those 300 million; I am one of those 300 million. I just took the placement test on Duolingo before this interview. I realized my Spanish is not quite as good as I thought that it was.
Luis: Very common.
Simon: But part of the learning process is actually realizing how good you are. And you have some software component that learns along with the person trying to learn to assess if they’re progressing.
Luis: Yeah, I mean a lot of the innovation with Duolingo has been to have an artificial-intelligence system in the background trying to really teach you better and what’s been amazing is just how much we’ve been able to improve it based on the fact that we have so many users. So for example, when we first launched we didn’t know whether we should teach plurals before adjectives or adjectives before plurals. And in fact nobody really knew what was best for any language.
But after we launched, we have so many users that nowadays if the system wants to figure out whether it should teach plurals before adjectives or adjectives before plurals, it just runs a test with about 50,000 people. So for the next 50,000 people that sign up, which takes about six hours for 50,000 new users to come to Duolingo, to half of them it teaches plurals before adjectives. To the other half it teaches adjectives before plurals. And then it measures which ones learn better. And so once and for all it can figure out, it turns out for this particular language to teach plurals before adjectives, for example.
So every week the system is improving. It’s making itself better at teaching by learning from our learners. So it’s doing that just based on huge amounts of data. And this is why it’s become so successful I think at teaching and why we have so many users.
Simon: Successful indeed. Duolingo is just voted the top iPhone app that’s available on iPhones right now. So definitely very successful. A lot of people too. Back to the teaching about.. or learning from previous things. You also previously co-founded CAPTCHA. And the esp test which is the lead required by Google. This is the image classification of which of the following has a stop sign in the picture and stuff. How important is image classification in that kind of work to the future of artificial intelligence and deep learning a lot of the stuff that’s going on here in MIT?
Luis: Well, I mean being able to understand images is one of the things that humans are really good at. You take a five-year-old and they can really understand what’s in an image or what’s in a scene. Computers are getting better and better at it. They’re not quite yet as good as humans. But I mean this is one of the biggest problems in artificial intelligence is being able to do that. I’ve pivoted to language. This is why I did Duolingo. But I mean, language is another one of those super important things in artificial intelligence that I think as soon as computers are able to understand language as well as humans can, we’ll be in a much more, much different world.
Simon: Where do we stand right now as far as computers’ or robots’ being able to understand things? You talk about human computation. A lot of the people are calling it crowdsourcing. But it’s kind of combining the best of humans and computers to do things, sort of figure things out. We keep hearing that they’re getting smarter and they’re getting better. Where do we stand in this?
Luis: Computers are getting smarter. … Not only are they getting smarter, I believe one day they’ll be as smart as humans. It’s pretty clear. Now it is impossible to tell when. I don’t know when that will happen. Probably not in the next five years. But we’re in a funny spot.
There are many things computers are so much better than humans at, like multiplying numbers or whatever. But there are some things that humans are really good at that even your five-year-olds are good at that computers cannot yet do. Like understanding really what’s in a visual scene or in a video or something. Computers can give you a basic understanding of it, but they can’t really tell you, “Oh, that’s a video of the president saying this and it’s in this context.” Computers can’t quite do that yet. But they’re getting better every year. It’s happening. … In the case of Duolingo what we’re trying to do is we’re trying to make a computer that is as good as a human teacher teaching. We’re not yet there. I mean human teachers are still better than computers.
But wouldn’t it be nice if we could have a computer that could teach you that is as good as a one-on-one human tutor? This is what we’re trying to do with Duolingo. That’s what I’m most excited about.
Simon: Sure. And our audience is individual investors. Maybe familiar with the space. Definitely not experts in this space. But where are a couple of things that you would say we should keep an eye on in human computation, in AI? What are a couple things that interest you that we should be watching?
Luis: Well, it’s a lot of talk about self-driving cars. I unfortunately think they’re a little farther away than most people thought a couple of years ago. It’s interesting to watch but I think it’s going to take a little longer to really get a full self-driving car. I’m very interested in just advances in how well we can teach. And the reason I’m interested in that is because most countries in the world have really crappy educational systems.
And in the case of Duolingo for example, we’re only 20% of our users are in the United States. Eighty percent are outside of the U.S. And most people are using it to learn English. And I think we make a huge dent in terms of how well we teach throughout the world if we can have an artificial intelligence that is as good as a human teacher. So that’s what I’m watching out for. That’s what I’m doing. And that’s what I will be looking out for.
Simon: Sure. And also in terms of human teachers, I know you are also a professor. At Carnegie Mellon in Pittsburgh. What are some of the things that your computer-science students are interested in right now? Are there certain topics that they really seem to be gravitating toward or certain topics that you’re teaching?
Luis: A lot of people are very interested in artificial intelligence. A lot of people are very interested in artificial intelligence for good. Trying to use AI for either solving things like climate change or trying to figure out how can we help people to live better lives? So it’s a combination of what the millennials want and artificial intelligence. Yeah, I think that’s what I would say they’re most interested in.
Simon: Very good. Well, Luis von Ahn and the co-founder and president of Duolingo. You can check out their software for learning different languages around the world.
Luis, thanks very much for the time today!
Luis: Thank you for having me.
Simon: Thanks for tuning in. Until next time.
John Mackey, CEO of Whole Foods Market, an Amazon subsidiary, is a member of The Motley Fool’s board of directors. Suzanne Frey, an executive at Alphabet, is a member of The Motley Fool’s board of directors. Simon Erickson owns shares of 2U and Amazon. The Motley Fool owns shares of and recommends Alphabet (C shares) and Amazon. The Motley Fool recommends 2U and Marriott International. The Motley Fool has a disclosure policy.