Thought Leaders in Artificial Intelligence: John Price, CEO of Vast (Part 6)

Sramana Mitra: Based on all the things that you’re doing, what are the trends and where do you see open problems that would be good pointers for new entrepreneurs to start companies in? If you were starting a company today, where would you start the company?

John Price: I can tell you what not to do probably much better than I can tell you what to do. After Shell Research, I ended up at Neuron Data. We were a C-based expert systems shell.

Sramana Mitra: I remember Neuron Data.

John Price: They recruited me out of Shell Research. The AI winter hit in the late 80’s. I partnered with some Stanford students out there and we create an AI company in the 90’s called Trilogy. I had a 10-year run with Trilogy where we built a $250 million AI company. We never mentioned the word AI. It was just tech.

We went through that and figured out how to make that work. Now, I’m in Vast and I am fortunate to have built quite a network of talent down here in the late 90’s. I recruited them all back to Vast. We have the same AI team that built the configurator. The lessons along the way are never should you think that AI is the center of the universe. There was this paper called Copernicus. It said, “AI is not the center of the universe.”

His point was it still requires great software engineering to deploy systems and products that people will use. I deployed an expert system at Shell in Lisp. People would argue with me. They kept telling me about objects. I said, “Those aren’t objects. Those are frames.” The great news about that is, you either learn a lesson because you read what somebody else said or you learn it the hard way. We learned it the hard way trying to deliver Lisp applications. You have to build products that solve people’s problems. AI is and will continue to be a great piece of the technology. I am not predicting another AI winter. I think that the viability, market, and the money is real. The results are real.

Sramana Mitra: The thing that is really different at this point is the level of data that is out there.

John Price: There’s no question. I built an expert system for helping blow out prevention of offshore oil rigs. I built 126 knowledge bases. The only reason I remember that number is because it was so many. 126 knowledge bases with tens of thousands of rules in order to make a recommendation. We were going to encode the knowledge to be able to emulate human expertise. That turned out to be not a very good idea. It was hard to maintain.

You have to give Google a lot of credit for helping usher this wave in. What Google discovered is, if you have lots of data, you actually can have reasonably unintelligent algorithms that can derive incredible knowledge out of that data. The data, in fact, is the gold. Here’s the gotcha. This is my next warning. It’s not the machine learning algorithm. I know everybody thinks, “I have a better machine learning algorithm.”

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