AI Is Not Eating the World (yet)

There is a growing disconnect in the tech world between people who write about Machine Learning/AI and actual practitioners in the field. For instance, take a quote recent techcrunch article that's been making the rounds:

In fact, it now appears that we will be able to achieve Artificial General Intelligence (AGI) some time around 2025. Technology is clearly expanding at a faster and faster pace, and, by many accounts, most of us will be caught off guard.

Or take this quote from a recent WSJ op-ed on Universal Basic Income:

The list goes on, and it also includes millions of white-collar jobs formerly thought to be safe. For decades, progress in artificial intelligence lagged behind the hype. In the past few years, AI has come of age.

The journalist in the WSJ op-ed cites Google's Go-playing AI as example that we've "come of age." The techcrunch writer cites a Ray Kurzweil book from 2012. AI's have been beating gaming machines for years - remember Deep Blue? Remember Watson? These AIs are cool and surely have replaced some workers or at least are trying to, but somehow we're still pretty close to full employment. What's going on?

Let's survey what the big tech companies are doing in the space to get a good idea:

• Facebook - FBLearner Flow - facebook has built an internal tool where engineers can come in, plug in datasets, train models and deploy them into production.
• Microsoft Cosmos - Before the facebook tool existed, Microsoft built something called Cosmos, which seems almost identical to fblearner flow. Engineers can plug in datasets and models to build and deploy machine learning.
• Google - Tensorflow - Basically a really dynamic platform for building and training all sorts of models, particularly optimized for GPU intensive tasks such as neural networks.
• Amazon Machine Learning - Basically fblearner flow/cosmos for the general public (and they're working on integrating it with iot)
• Uber - Job Description - looks like they're building something similar to Cosmos or fblearner flow. Not a surprise given uber is populated by tons of ex-facebook folks.
• Amazon's popular virtual assistant, Alexa, has gotten a whole smarter in the past six months. The tech giant boasted on Friday that Alexa now has 1,000 skills.
• airbnb - Aerosolve

What do these all have in common? They're all tools for humans by humans to automate domain-specific workflows. I am not convinced that we are anywhere near having a general "AGI" machine learning tool that can get around this. At least for the foreseeable future, we need engineers and domain experts to make this all work.

Who are the winners in this world?

• Big Tech companies that already have the machine learning workflow technology and can extend to more general cases. See above list.
• Cloud Tech companies who can build and support the layers required for this to work. I wouldn't be surprised to see AWS drop some big products in this space before November (AWS Machine Learning is somewhat technically disappointing at the moment)
• Consultants who find human workflows and continue the process of automating workers away by implementing some form of Machine Learning. Note, this isn't all that different from what has been going on for the last 40 years where consulting firms have sought to automate expensive workflows for big clients (remember the movie Office Space?) This is just another tool in their arsenal (which includes things like offshoring, robotic automation, and so on)

Unless there is a serious breakthrough in extending Machine Learning tools to domain specific fields, I don't yet see a world where the AI can eat everything without a lot of human intervention (yet).

Published by:

Thomas Hayden