Wednesday, November 23, 2016

Big Data, Ed-Tech and Precogs


Reading Kevin Kelly’s book (The Inevitable) motivated me to rewatch the 2002 movie Minority Report. The movie, directed by Steven Spielberg, is based on a short story by Philip K. Dick. The year is 2054; an experimental unit named Pre-Crime has been operating in Washington D.C. for six years and has reduced murders to practically zero. How did they do it? They have precogs, individuals who have extrasensory perception (ESP) skills allowing them to see murders before they inevitably take place. The precogs are a “hive mind” linked to computers and data, and as one character says in the movie: “Don’t think of them as human.” They rest floating in a pool with as little to disturb them as possible. They are fed nutrients and their cerebral activities are linked to Pre-Crime’s sophisticated computer system.

For a 2002 movie envisioning the future, Spielberg did a fantastic job. The featured tech includes a combination of things I’ve written about recently: virtual reality and augmented reality, connected to and drawing from a huge database (the Internet!). Big data is combined with the visual streams from the precogs, allowing the main protagonist played by Tom Cruise to conduct a virtuoso investigation. As he waves his hands in the air, different streams of information are constructed and displayed for the team. It’s like watching a maestro conduct an orchestra. (The Iron Man movies borrowed heavily from this imagery.) While we are still a long way off from 2050, we are well on our way to developing that sort of tech.

The precogs provide one stream of data into a huge computing database, that allows pinpointing the day, time, location, and actions in a future event. One way to think about this is that with enough data, someone observing past behavior could make reasonably good inferences as to future behavior. If you didn’t do well on a pre-test, skipped class, didn’t take notes, didn’t do the homework, didn’t study, then you’re likely not to do well on the exam. Humans connected to the internet be it via wearable devices, mobile phones, or sitting at computer terminals, are constantly giving up more and more of their data to this database. A sufficiently strong A.I. just might be able to piece together some aggregate future behavior based on past behavior in comparison to a huge database from hundreds of millions of individuals. This A.I. may even be able to assign a probability value from the pre-calculated statistics.

This past week I received yet another spam e-mail from ed-tech. I usually delete these without reading it, but since I had been pondering the precogs, I decided to read the e-mail. This one was from Pearson and was advertising its adaptive learning platform, Knewton, integrated into Mastering Chemistry. I have some idea of how these systems work (I’ve mentioned Knewton in a previous post); I read some of the initial literature surrounding ALEKS before it became proprietary. Earlier this year I also read a white paper (can’t find it now) taking stock of where we are with adaptive learning systems in higher education.

I watched the short embedded video advertising Pearson’s product. Not surprisingly, I didn’t learn anything new from a technology standpoint. However what was more interesting was the sales pitch. The A.I. system gauges your starting point by asking you questions (you solve chemistry problems) and based on your mastery or lack thereof, the system shunts you along particular pathways that are personalized for you. They have been personalized based on an ever-growing training set – the big data and the analytics. Every choice you make, key you press, even the time you take working on a problem, all that is added to the database. If there was an eye-tracking system, I bet they’ll be gobbling up that information too. In Minority Report, that’s how advertising works. As Tom Cruise wanders through the subways and the shopping mall, ubiquitous eye trackers scan him, personalize his ads, and make suggestions.

It’s not a big stretch to “classify” students (based on their performance in these adaptive learning systems) into categories of academic competence. Pearson and other companies are already supplementing other parts of student tracking and advising (this is a major push) and selling their products to the Student Affairs part of the university. They are probably going to make inroads into Career Services as part of their encompassing strategy (maybe they already have!). Soon the system will be helping students sort through potential careers based on academic potential, extracurricular activities and any other information it will happily gobble up into its big data analytics algorithms. With more data, it can assign probabilities that a certain individual will end up in a certain career. (It can probably already assign an exam score probability.) Or make a probabilistic prediction that an individual will take certain actions. Starting to sound like a precog yet? A hive mind, perhaps?

In her third year at Hogwarts, Hermione elects to take Arithmancy – mathematics that predicts the future. How about algorithms that predict the future based on big data? I hereby invent the word Algorithmancy. (You heard it here first!) I’ve speculated that Advanced Arithmancy is akin to theoretical Physical Chemistry. Algorithmancy would be what I do as a computational chemist – I apply theoretical models in physical chemistry to make predictions on how ensembles of molecules might evolve taking into account their intrinsic “chemistry” but also the “environment” they are in. I happen to study the origin of life using computers, but might I also be originating life in a computer? We call it artificial intelligence. But is “artificial” the right word? Maybe it’s just a different kind of intelligence, one that we don’t understand and seems alien to us even as it grows exponentially feeding on vast networks of data. Alien intelligence may be the new A.I.

No comments:

Post a Comment