Thursday, July 30, 2015
Assuming catastrophe doesn’t strike, the blog will be back on Monday, August 10, once again reporting from New Jersey.
Once had a real estate transaction nearly derailed because the buyer cashed her paycheck and deposited the cash. Her lender demanded she prove where the cash had come from.
The error in the article is that it confuses the cost for educating a good student (one who learns at the expected rate) with that of a student that repeats many classes. The innovation mentioned in that article is not "technical" (the author is clearly aware that the gains from technical innovation were obtained 40 to 50 years ago with larger classrooms and film or video recorded lectures) but pedagogical. Increasing learning at the same cost -- via more effective teaching -- will reduce some signficant fixed costs of a college and the net cost of an education. This is a valid point even if you don't believe those same learning gains can come from the next generation of teaching machines. However, the same benefits can be found from grade inflation, so it might not show up much unless you compare outcomes assessment with cost.
Whether learning can be done without any humans at all is the key question, and why you need to think quality rather than kind when it comes to "technology". I will mention that the investment side of this issue is whether a cheap IBM Watson computer combined with Pearson's learning data can replace people to a degree that previous innovations (starting with programmed learning and then PLATO) did not. However, on the cost side, the more signficant question is whether the cost of this will be lower than (say) video lectures plus grad TAs and, even more importantly, whether the difference will show up in the consumer price of the goods. There are pickup trucks today that are priced like they were made by Italian craftsmen rather than robots!