Analyst firm IDC released a report predicting that spending on artificial intelligence systems will be $12.5 billion in 2017
and more than $46 billion by 2020. Of that $12.5 billion this year, more than one-third ($4.5 billion) will be on applications, for use cases such as threat detection, fraud analysis, public safety and pharmaceutical research. They will represent an even larger overall share of AI spending by 2020.
This is a long way of saying that AI is mostly a feature, not a market unto itself. If you’re selling software for, say, drug discovery, you’re in the drug discovery business. AI, and any other algorithms and technologies you utilize, are just means to that end.
Maybe you saw the news out of Stanford yesterday that researchers there have developed deep learning algorithms to aid in drug discovery
, and released their code as the DeepChem library. It might have applicability in other realms of molecular chemistry, but nobody is going to use it for robotics anytime soon. If it’s commercialized, it will almost certainly be sold to pharmaceutical companies.
Applications, devices and other “AI-inside” products are where most of the money will be made. If I’m a CISO and IBM tells me, “Use Watson,” I might laugh it off. If IBM says, “Use our Cognitive SOC, powered by Watson,” then perhaps my ears will perk up. Buyers don’t buy AI, they buy things that are useful for their jobs in their fields.
I guess what I’m saying is that while it’s important to track the effects of AI in various markets, we don’t want to lose track of the markets themselves. AI is a tool to make other products and systems betters, and at the end of the day it’s those products, systems and the fields they’re advancing that really matter.