#mlconf

There has recently been an explosion of cynicism in the Twitterverse (and generally in the tech community) about machine learning and AI. This was taken up a notch after Facebook’s F8 conference earlier this week and the splashy launch of their bot platform on Messenger.

I understand the cynicism. When you dig in, many of the companies claiming to be leveraging machine learning are doing nothing of the sort. It’s easy to find companies that are building weak ML-driven user experiences, turning off early adopters. Others are building products centered around gimmicky ML features that won’t create any sort of long term barrier to competitors, frustrating would-be investors.

I think we’re probably on our way up to the peak of the AI hype cycle right now. A lot of money is going into projects that will not end up being valuable to society. It's normal at this stage, and it’s just fine.

Let’s get past the hype. Bots are an interesting use case for machine learning, but it goes much deeper than that. There is fundamental innovation happening around us. 

I spent most of the day today at MLconf in New York. It was a packed house at 230 5th Ave — a melting pot of engineers, scientists, and entrepreneurs from every corner of the globe. As a physical scientist / engineer, I usually feel like a techie. Not today.

Consumer-facing bots came up a few times during the conference, but they weren't the focus. People were talking about new, advanced statistical techniques that I had never seen before in my research on the space. Software to allow data scientists to easily scale machine learning algorithms to parallel computing using only Python. Neural networks for advanced speech recognition, trained off of unimaginably large data sets. Models to predict medical issues from streams of live patient data from hospitals. I got a demo that blew my mind... but I promised not to talk about it publicly, so I won’t.

Here’s the simple truth: many of the smartest people in the world are throwing every ounce of their brain power into creating the building blocks for a machine learning software ecosystem. There have been some success stories so far, as well as some failures. While it’s still early, I don’t think it’s ever a winning proposition to bet against humanity’s ability to come together to solve big, technical problems.