- David Cheriton, an early investor in Google, says self-driving cars, AI, and IoT are overhyped.
- One area that Cheriton doesn’t think is on its way out is automation.
- He said he liked Darwin’s view that “most innovations are not very useful or compelling.”
- Visit Insider’s homepage for more stories.
In September, Forbes contributor Bernard Marr listed artificial intelligence and vehicle automation among the five biggest technology trends to keep an eye out for in 2021. Other industry analysts, such as the global research and advisory firm Gartner, have also included AI on their lists of top strategic technology trends for this year.
But one industry expert who doesn’t agree is David Cheriton.
Cheriton is best known for being an early investor in Google. But he has many other feathers in his cap, including being an early angel investor in VMware and the founder and CEO of Apstra, which helps companies simplify and automate their data-center networks.
One area that Cheriton doesn’t think is on its way out is automation; in fact, he said he believed that the potential of computing and automation was still being underestimated.
“The biggest game in town is still the biggest game in town and will be for as far as I can see,” he said, adding he once gave a talk where he claimed basically everything that could be automated would be automated.
“That was my first rule of law,” Cheriton said. “The second law was that everything can be automated.”
He added that this was where he departed from those who push AI’s ability to duplicate what human beings can do as a big, interesting challenge.
“We make so many nutty decisions and can become our worst enemies through human error,” Cheriton said. “So what’s not to like about living in an automated zoo where we are taken care of in a much better way at a much lower cost and have a higher quality of living?”
He also thinks people overestimate the degree that it’s a threat to human beings.
“Real modern computing is about 70 years old, more or less, but the situation for the world has been essentially the same as it’s been for the recorded history of human beings,” he said. “With that in mind, our biggest threat is other human beings, not the tools we build.”
Cheriton believes there are more “down and leaving” technologies that people are paying too much attention to, he said.
“I like the Darwin view that most mutations are harmful and most innovations are not very useful or compelling,” Cheriton said. “I look at that to see if something is ‘down and leaving.'” He pointed to what he said were three overhyped technologies to be wary of:
A lot of money is going toward self-driving cars. According to a recent article in International Business Times, investment in the autonomous-vehicle industry is more than $100 billion. But Cheriton believes they’ve run their course, he said.
“I’m not saying self-driving is impossible, but some of the AI-focused approaches aren’t really applicable,” he said. “Saying that 99% of the people get to their destination without destroying their car is not acceptable. It’s a challenging technical problem, but the problem needs to be understood a little bit better. Then you shape the solution.”
He said there were multiple search engines that existed before Google that used various techniques — but the competitors didn’t solve the problem of having users trust the results.
“There were other search engines that tried to provide incorporated semantic understanding into this, and they didn’t survive,” Cheriton said. “Google figured out that they needed to solve the trust problem, not build in a lot of semantics directly. That’s how you end up with a winning solution.”
He acknowledged that he was unsure how close companies were to identifying the right approach with autonomous vehicles but said he thought they were getting closer. With that in mind, the investor “can’t just ask if they are betting a winning horse,” he said — they need to understand how long they have to feed the horse until it gets across the finish line.
“Some people have speculated that even Waymo might be 10 years away from really being able to say we have autonomous driving,” he added.
Cheriton said AI was “sort of an interesting area” but that it was way oversold for what it could accomplish. He said that when you strip away the terminology, machine learning is basically statistical regression with significant improvements on what statisticians have typically done.
“There are cases where AI improves processes, but then there are other cases where I think they’re barking up the wrong tree,” Cheriton said. “And at some point, reality hits and people stop investing there.”
As an example, he said the application of AI to self-driving cars was misguided.
“We’ve seen that a lot of investments go there,” Cheriton said. “And it’s much harder to get anybody to invest in that direction at the startup-company level, so the hype cycle is starting to run its course.”
The Internet of Things
Cheriton also believes that Internet of Thing’s popularity is on the wane.
“IoT was sold as something where you were going to have all these devices that you’d want to connect to the internet,” he said. “Therefore, there’s this big need for all this technology associated with that.”
Whenever Cheriton hears people pushing something based on technology, he’s suspicious, he said.
“You find in some of the literature on IoT that your fridge will be connected to the internet, your toaster will be connected to the internet, your washing machine will be connected to the internet,” he said. “Yet at the end of the day, people are going to be asking, ‘What would I do with them? What’s the benefit if I have to pay extra to have them all connected to the internet?'”
He added: “When the market is very small, there still might not be enough support to do something. But if you identify a large market, then there’s a chance for success. IoT still needs to get there.”