Thought Leaders in Artificial Intelligence: John Price, CEO of Vast (Part 7)

John Price: In order to solve a problem that is worth solving with machine learning, the biggest problem is the amount of data you have to aggregate, normalize, and get in shape. Once again, it’s not the intelligence. AI is not the center of the universe. The problem is shifted from software engineering. You pick your vertical you want to solve and get that data in shape for your machine learning algorithms. Be careful because you’re underestimating the amount of effort to get the data.

Sramana Mitra: What you said is exactly right. We keep harping on this point. To set up something that would work in an AI model with machine learning, you really need to understand the domain in which you’re building the problem.

Each domain is different. The setup that you need to apply AI comprehensively to each of those domains is different. Domain knowledge is becoming far more valuable and important in making AI work than anything else.

John Price: I agree with you 100%. It’s only because we have 10 years in this very narrow segment that allows us to unlock the value in that data. I couldn’t agree with you more. The reason is very explainable. They think of these solutions in the horizontal world. Steve Case did us all such a service about all these other verticals where we’re going to go apply these horizontal technologies. He does a really good job of laying out everything I’ve experienced in the verticals as I’ve gone deep in these verticals. The thing we haven’t talked about is you got to build out all the partnerships to get all the data.

Sramana Mitra: That’s why I asked you what your data sources were. A lot of data is behind the walls. What is the case for them to give you that data? What you can do with it is a technology and architecture issue but then there’s all kinds of business development questions that you have to face. Why would they give you access to that?

John Price: You now I’m smiling. I wish my Board appreciated everything you said.

Sramana Mitra: I did a couple of AI startups early on in the 90’s. You did it even earlier than I did. Even in the 90’s, it was way too early.

John Price: I’m the glass is overflowing guy or I wouldn’t have been in AI for 25 years. There’s never been a better time than today. The two biggest things is the availability of data and the economics of data. Without things like Hadoop, we couldn’t afford to store all of the data that we’re keeping around. Now we’re a profitable multi-million dollar company. Startup years are thin. How are you going to store all that data? All of that is solved.

Second, it’s all about the fact that today, everybody has a computer in their pocket than when you and I were building out AI systems. It’s actually orders of magnitude more powerful than the systems we were trying to deploy it on. It’s ridiculous how much computing power we now have.

Since we do so much analytics and insights around the data, our performance is affected by having to go back to the server all the time. We just built a native client implementation of our full stack so we can put in your Android or iPhone, 500,000 full descriptions of automobiles as well as all my analytics, and do all the processing on your phone. Our application runs 100 times faster. Now we’re building applications more like vertical Tableau for automotive. I can compute loans on every car in the country simultaneously. All of this because of the power in those phones today.

Sramana Mitra: When I think back on those days, the apps were all fragile and very unscalable. Great conversation. Thank you for your time.

This segment is part 7 in the series : Thought Leaders in Artificial Intelligence: John Price, CEO of Vast
1 2 3 4 5 6 7