Monday, May 30, 2016

The Case for Humans

It may seem strange to a lot of people that Zeal has put so much focus into live math coaches.  After all, isn't software supposed to be eating the world?  Isn't artificial intelligence supposed to expose us humans as insignificant?  Maybe some day, but in education, human relationships are central, and computers are terrible at that.

So why wasn't that obvious to me from the day I left Rocketship to start Zeal?  Our original thesis at Zeal was that if we did a better job at providing teachers data on their students, that would help them spend much more personal attention on students, helping teachers meet students where they were.  And that did happen.  But it didn't happen at the scale we wanted.  About 20% of the 2200 or so Zeal assessment classrooms showed significant increases in personal attention, 80% did not.  So about 18 months ago, we went to talk to our power users, many of whom were very enthusiastic to collect data on their students, and we asked them "Why aren't you using this to differentiate more, to provide more personal attention?"  And the answer was "We are really trying, but there just aren't enough hours in the day."  That's when we realized that data was not going to solve teachers problems, giving them more hours in the day was the solution they needed.

We went back to those teachers about three months later and said, "What if >we< used the Zeal data to fill the gaps your students have?"  To a teacher, everyone said, "Yes please!  I feel so guilty that I can't help everyone.  If you could do more, it would be huge for me."  That's when we started coaching.  We toyed with the idea of trying to fill those gaps through software as everyone else had, but all of my experience doing tutoring at Rocketship told me that when a student is frustrated, when they don't understand something, computers are exceptionally bad at connecting with them, motivating them, and seeking out the issue.  Tutoring at Rocketship was phenomenally successful, but far too expensive.  At Zeal, could we figure out how to create a new role, a coach, that would work with a classroom of students, intervening like a tutor, but doing it at the same cost as computer-based instruction?  If so, we could help our teachers provide more personal attention.

So we went out and found a lot of retired educators and started connecting them to students online with a shared audio connection and white board to work on problems together.  It was magic from the first session.  Students felt heard, they re-engaged, coaches told us that this was the best job they had ever had.  It was a pure student-teacher interaction, and it worked, generating learning rates 10 times what computer-based instruction generated at Rocketship when I was there.

Make no mistake, augmenting teachers with coaches is a lot more complicated business than just writing software.  But what it takes advantage of is the power of letting coaches work anywhere at any time.  There are so many more people that are able to commit to that, than the twelve hour days of managing a classroom of students.  And many of those people are really good at math and really love kids.

So will we see a future where there is more automation of instruction?  I'm sure we will.  But I also know that along with art and poetry, the critical role of the educator working with a student when they are confused or frustrated, is going to be one of the last to go to the robots.