A small team of student AI coders beats Google’s machine-learning code

The Fast.ai algorithm was trained on the ImageNet database in 18 minutes using 16 Amazon Web Service instances, at a total compute cost of around $40.

Source: Knight, W. (2018). A small team of student AI coders beats Google’s machine-learning code.

A few things that I took from this:

1. Access to massive data sets is still driving the progress of AI, especially with respect to machine learning.
2. But, as algorithms improve we’re seeing small teams able to do more with less.
3. Which means that big players like Google, Facebook, Amazon, etc. may not be as dominant in the future as they appear to be today.
4. Or, maybe they’ll just buy up the smaller teams.