As hot as AI is, and as powerful as ML models can be, some important pieces are still missing without which AI risks running afoul of minimal standards in rigor, accuracy and, indeed, ethics. Here’s a quick inventory:• The internet of things (IoT) has made a large variety of constantly-updated time series data available. • Cloud computing has made it affordable and straightforward to requisition sufficient processing resources to build the ML models using the entirety of such datasets without resorting to sampling. • Rapidly advancing graphics processing unit (GPU) technology is reducing training times for new models and inference/scoring times when using the models in production. The cloud makes the latest GPU technology available in abundance on an elastic basis.

Source: Forbes May 24, 2018 13:30 UTC