Monday, April 10, 2017

Will Investors Love Fake Teachers?

Well, this was, of course, inevitable. We've heard a lot about CGI students as a training (and evaluation) technique for teachers, a freakish notion in which teachers or neo-teachers square off against cyber-puppets in an uncanny valley showdown.

And we've also hear endlessly about artificial intelligence being used to craft a personalized education program, somehow.

I'm from the future and I'm here to teach you stuff

So why not extend both ideas. Alex Salkever is here with the pitch, not in an education site, but at MarketWatch, a site for investors searching for the Next Big Thing. Salkever is VP of marketing/communications at Mozilla, and the co-author of The Driver in the Driverless Car. On LinkedIN he says that he helps "companies grow by building communities and creating memorable content."

And he thinks avatars are the teachers of the not-too-distant future.

Do avatar teachers seem too far afield? Let me ask another question: Does your child use artificial intelligence to learn? More parents will be answering yes in the coming years.  

It's easy peasy. "AI covers everything from smarter automated robotic reservations systems for airlines to tiny food delivery robots rolling through Washington, D.C. to virtual pharmacists that spot potential for adverse drug reactions based on our past histories and current prescription regimes." And how much harder than that could teaching be? Well, actually, the avatar teacher would have to be "strong AI," and "we aren't at Strong AI yet," but don't worry--

...we already can educate all children much better using today’s technology than with more traditional methods in the classroom. That’s because computers provide reliable feedback, don’t get tired and can guide learning to emphasize areas where reinforcement is needed. 

Sigh. Yes, it's one more technician who's sure he knows how to replace teachers even though he doesn't understand the job. Or hasn't done enough research to know tech-heavy models like Rocketship Academy (motto: just keep your young eyes on those computer screens) have failed to launch. Or that tech-heavy models like virtual schools, aka cyber schools, have been shown to actually be worse than nothing at all.

But while Salkever acknowledges that some of the necessary tech is still emerging, he has one piece that he thinks is key-- virtual reality goggles.

I won't sell you the data-- not even to save the farm

The CGI teacher will be right there in the goggles in 3D! And the goggles will read all manner of physiological feedback from the student. So if you thought Big Data's collection of test scores and personal data seemed scary and excessive, just imagine data miners scooping up your child's every physical reaction to every stimulus provided by the programming. Salkever knows this is hugely Big Brothery, but parenthetically, he hopes "that we will build proper privacy and data control mechanisms to let pupils and parents decide who can see their vital information about learning and biological responses." Yes. Because keeping data secure is something that we've been excellent at so far, and because corporations regular say, "Yes, we could sell that to you, but it would be wrong, so we just won't." And remember-- this is a pitch in a site aimed at investors.

Not creepy enough for you yet?

The world becomes the classroom and the classroom becomes the world. This isn’t to say that the real world goes away. To the contrary. The blending of the two, with our guide to teach us along the way, creates a seamless digital and analog learning space.  

Wow, that's-- wait! What?

Look, there are so many things wrong with Salkever's pitch. Let's just pick a couple.

1) His fake digital teaching is only good to the extent that it resembles teaching done by real humans. So how is it better? His suggestion is that it doesn't become tired or get bored or ever miss an answer, but human students do all those things, so why is a piece of software that doesn't well-suited to teaching them. We've got research right now that tells us that students learn better when they see someone like them in the classroom. Right now we're drawing the conclusion that this means black students do better when they have even one black teacher; I'm betting we can learn that human students learn better from human teachers.

2) Software is only as smart as the person programming it. Avatar teachers will not be programmed by God, like digital manna downloaded from heaven. They will be programmed by humans. That means the artificial intelligence will not be any smarter about, say, American literature than the person who programs them. And if you think everything that can be possibly known or understood or divined about American literature can be loaded into one program, you are probably not a very good literature student. The possible responses to a chess move are many, but they are finite-- it is possible to program every possible situation into one digital brain. The possible responses to all the works in the American literature canon is infinite. Which brings me to

3) A digital teacher depends on a "fill the pail" model of education. It assumes that teaching is about downloading pieces of information from your brain to the students'. It assumes that there is no discovery, no exploration, and definitely nothing to learn in a classroom that is not already known. In short, digital teachers are the perfect tool for replacing teachers from, say, the late 1800s.

4) The people who are pushing this stuff are largely ignorant of teaching. Salkever allows that there will not be teachers any more-- just "lots of very good coaches (once called teachers) who focus on the creative, motivational and communications aspects that are far harder to translate into the digital realm." In other words, teachers. Because the creative, motivational, and communication aspects are the largest part of the teaching game. Presumably the "coach" would also handle things like breaking up fights and interrupting games like "Who can shoot a spitball closest to the avatar's nose."

I suppose those who believe that education is a simple technical task could really close the loop here and put a digital teacher avatar in a classroom with digital avatar students. It would not involve any actual humans or human relationships at all, nor would it have anything to do with education. But, oh, the investment opportunities.


  1. Salkever needs to spend one week teaching an over-crowded 7th grade class in a high needs school.

  2. Put this together with Rahm Emanuel's most recent stroke of genius and you get:

    When I graduate, I want to be a digital teaching avatar!

    Jeebus wept.

  3. This is just beyond lunacy. For example: 15 years, hundreds of thousands of kids, 195 influences in John Hattie's research, and I'm guessing the brilliant cutting edge Salkever would say, "Who?"