Gigaom AI Minute – November 15

Transcript

How can enterprises deploy artificial intelligence today? I think it's kind of a three-phase process. The first phase involves using someone else's data and someone else's technology to draw new kinds of conclusions from it. Beyond that you would use packaged technology along with your own proprietary data and study that and see what you can pull out of it. Then finally you will be at a level where you build your own technology and power your own data and do highly customized solutions.

Every artificial intelligence project, at least at this point, really should be measurable, at least in some way, demonstrate a quantifiable improvement in some metric. I mean, you're using data to train it so presumably the same kinds of data can evaluate the outcome. Artificial intelligence projects for enterprises need not be particularly large. Small gains are often the lowest hanging fruit as this technology deploys, and the number of places that you can get those small gains is quite large.