Thought Leaders in Online Education: Amesite CEO Ann Marie Sastry (Part 2)

Sramana Mitra: Double-click down one more level for me and tell me how.

Ann Marie Sastry: So, we do this by selecting articles, ranking, and then training algorithms to select articles that are closely related either by hand selection or by other means. 

Sramana Mitra: What are the sources of content that you’re applying algorithms on?

Ann Marie Sastry: That’s key. Anybody can go to Reddit or Quora, but most universities and college already take license for large databases of articles that are peer-reviewed that often don’t see light of day.

The irony is that colleges and universities have a huge storehouse of information that very rarely gets used in graduate and undergraduate courses. Additionally, they have qualified journalistic sources. It may have an editorial bent but the reality is those are qualified sources of information. They adhere to journalistic standards that require them to use facts.

When you’re teaching an economics class, if there’s a question about GDP impact, that’s highly relevant. When you look at study after study of behavioral science on learning, it’s very clear that engagement actually precedes learning. You don’t learn and then become engaged. You engage and then you learn. Maybe we don’t really teach anybody anything.

I was a professor for 17 years. I was considered to be pretty good at it. I never thought I was that good at it. I always struggled to bring relevant content into classes. There are just physical limitations. For me to realistically read 10 to 15 articles a week was a lot, but our boss can read hundreds of thousands of articles a week.

The way we think about it is, we should be using the power available in automated algorithms and learning algorithms to cull information and bring qualified information.

One thing that teachers and professors can do very readily, after they achieve some level of expertise is, you may not be able to find the needle in the haystack. You may not be able to find the top 15 articles published on philosophy this week or on machine learning around logistics, but when you’re shown an article, you can use your expertise to comment on it and start a conversation on it. That’s very powerful for students.

That’s the first bucket. The second is to automate how people interact with a platform. The experiences are very sticky. For Netflix and Spotify, it’s so sticky. You’ve got a beautiful menu that’s built just for you. Education is not there yet. Why is it not there yet? The data from any experience, when it teaches a learning algorithm, can serve up a differentiated and customized experience.

There are a myriad of aspects of learning, such as the number of times you’re notified and how suggestions are made, that we learn on the platform based on the user’s behavior and try to adapt to that in a very data-driven way. Just as a side note to that, we have a very strong ethos that you shouldn’t have to trade your data to get an education. We do not allow third-party to look at data. We do not sell our data or advertising on the platform.

The third way is in looking at the discussions and suggesting content that goes where the course goes. Sentiment analysis can be very helpful in that. Behavioral science is very clear. It’s not that surprising. When you’re nice to people, they tend to do better. When you use encouraging language, people tend to  respond better. When you use supportive language, people respond better.

Using sentiment analysis to drive better outcomes by more friendly and more supportive interactions is the third way where you can really enhance a learning experience. That learning experience is really a conversation not only between teachers and students but also among the students themselves.

This segment is part 2 in the series : Thought Leaders in Online Education: Amesite CEO Ann Marie Sastry
1 2 3 4 5 6