AI in Education Still Learning New Tricks
Fat, drunk and stupid is no way to go through life, son. Those aren’t just the wise words of a fictional college dean in National Lampoon’s magnum opus, Animal House. It’s an apt description of today’s education system in the United States, which ranks a middling 38th in math out of 71 countries and 24th in science. It’s not for lack of money. The U.S. spends more than $600 billion on primary and secondary education, which works out to more than $12,500 per student. Technology to the rescue, right? Well, in 2016, the U.S. spent an estimated $160 billion on education technology, or edtech, according to a report by McKinsey Global Institute on the rise of artificial intelligence. So, is AI in education the silver bullet, Bluto?
Not anytime soon. Experts like those who write reports for McKinsey say the type of automation currently offered by AI platforms would have the least impact on education, right after activities like pet-sitting and curing male-pattern baldness.
However, a new report from McKinsey does see a bright future for AI in education on several fronts:
- Bridging the skills gap
- Attracting and retaining students
- Unleashing personalized learning
- Releasing teachers’ true value add
- Virtual teachers
It’s the third bullet point we want to look at more closely in this article. Personalization through artificial intelligence is becoming something of a theme lately. Such narcissism customization can help boost sales, for example. But we also recently discussed how this trend toward building an online experience tailored for each individual carries a certain danger of becoming an echo chamber, where your every belief and desire is simply reflected back at you. In education, however, people learn at different speeds and respond to different techniques. As the McKinsey report notes:
Artificial intelligence could improve adaptive learning and personalized teaching by identifying factors or indicators of successful learning for each student that were previously not possible to capture. In addition to monitoring such variables as the number of times a student pauses during a lesson, the amount of time needed to answer a question, and the number of times a question was attempted before getting it right, computer vision and deep learning could call in new information such as mouse movements, eye tracking, and sentiment analysis, delivering a deeper insights on the student’s performance, confidence, mindset, and cognitive ability.
In other words, AI will soon be able to do what any Chinese or Jewish mother already does today.
As investors, we want to know is there any money in all of this? Not really. At least not yet. Private investment in educational technology, defined by the McKinsey authors as the use of computers or other technology to enhance teaching, grew 32 percent annually from 2011 through 2015, to $4.5 billion globally. The rise of online education sites like Coursera and Udemy, which employ algorithms behind the scenes, are helping to democratize and personalize education so that people will make better choices than settling for an MBA. Skyrocketing higher education costs and increased competition for the best schools will drive more parents to pay for personalized education platforms for their kids.
In the second half of the article, we’ll look at a few startups making learning personal with AI in education.
Plotting a New Course
New York-based Knewton is one of a number of companies doing adaptive learning, basically designing coursework and tracking progress for each individual student. Founded in 2008, Knewton was at the forefront of adaptive learning and has amassed $157.25 million in funding from VC firms like Accel Partners and companies like Pearson Education, a British-owned education publishing and assessment service for schools and corporations. While many of these adaptive learning companies deal in big data to build their programs, only a few like Knewton explicitly say they use AI. From its promotional material, the company says:
Knewton adaptivity is distinguished by its scientific rigor and tremendous scope. Using advanced data science and machine learning, Knewton’s sophisticated technology identifies, on a real-time basis, each student’s strengths, weaknesses, and learning style. In this way, the Knewton platform is able to take the combined data of millions of other students to help each student learn every single concept he or she ever encounters.
Does it work? According to the company, after Arizona State University started using its developmental math courses, pass rates rose by 17 percent, course withdrawals dropped by 56 percent, and 45 percent of students finished four weeks early. However, Knewton itself has had a hard time making the grade of late. The website EdSurge reported last month that the company has been struggling. Its founder recently departed and Pearson, a major investor and client, is out for recess.
About 350,000 teachers reputedly use Front Row’s software platform to help students learn math, social studies, and other subjects. Based in San Francisco, Front Row has raised $6.6 million since it was founded in 2013. In math, for example, the company offers more than 30,000 math questions for students up to eighth grade. The kids start off with a diagnostic test, and then the algorithms go to work to tutor the lesson at each student’s own skill level.
Does it work? In one third-party study, students who used Front Row received test scores that were nearly 10 points higher than their peers. Those students were also two months ahead in math achievement. Forbes even featured the company a couple of years ago.
Thinkster Math, formerly known as Tabtor, just does mathematics. The Indian company, with an office in New Jersey, has raised $4.7 million, including a $3.7 million Series A last year. It provides real tutors, but employs AI and machine learning to visualize how a student is thinking while he or she works on a problem. This purportedly helps improve how fast and how well its tutoring staff can address problem areas. The program includes an app so parents can track their kid’s progress.
Does it work? We didn’t see any hard numbers on the Thinkster Math website, just some parent testimonials. Programs start as low as $60 a month but you get the Charlie Brown teacher who only says, “wah, wah, wah, wah, wah”.
A Swedish company called Sana markets itself as an AI company. It claims to use deep neural networks, which mimic the human brain, to optimize learning for each student and to predict how he or she will perform on future interactions. The algorithms then select the content that will result in the best improvements over time.
Does it work? We couldn’t find much on metrics or money behind this startup, except that the Stockholm-based company is backed by a “renowned set of investors and advisors”. It seems to be the most heavily invested in AI of all these startups in terms of its machine learning platform.
In the course of our research, we came across a few other AI in education startups that we thought might interest you:
Such as a talking dinosaur from Elemental Path, a New York-based startup that has raised $4 million and makes what it calls CogniToys. These cutesy Amazon Echo-type of toys connect to the internet and are capable of answering questions from children, such as, “Who is that strange man with mommy?” It’s powered partly by IBM’s Watson AI, according to CNET.
Gradescope gets an A for effort. It uses AI to help teachers grade papers en masse. It amassed about $2.5 million toward that goal in April 2016. Founded by teaching assistants overwhelmed at grading papers for a UC Berkeley computer science class, Gradescope is used in roughly 100 universities to grade 55,000 students. The company says automation through algorithms reduces grading time by 70 percent.
Lexplore (formerly Optolexia), another company based in Stockholm, has taken in $6.75 million to develop a test to identify dyslexia in children. Students read a short text on a computer screen while an eye-tracking device records the eye movements. Algorithms then go to work on the patterns to analyze if the child is at risk for dyslexia.
AI in education is one of the least mature examples of artificial intelligence we’ve come across in some time. Frankly, it’s difficult to assess just how much is AI and how much is just smart software with some of these companies. We certainly don’t expect to see teachers lose their jobs to machines any time soon.
Despite what the pundits say, FAANG stocks (Facebook, Apple, Amazon, Netflix, Google) don't give you real exposure to AI. Read about 7 stocks that give you true pure-play exposure to AI in our guide to investing in AI healthcare companies, freely available to Nanalyze subscribers.