He hasn't changed his tune on that issue, but he does express grave concern over the dismal consequences that kind of shift would have on the workforce.
Speaking on Tuesday before an audience of about 300 people at the MIT Technology Review Digital Summit in San Francisco, Khosla maintained that in most areas of the economy, machines can do better than humans.
For example, consider the fast food industry. Khosla, a partner at Khosla Ventures, said that his firm has invested in a company that created a hamburger app that allows users specify exactly how they want their hamburger prepared.
This reduces the need for a middle person, such as a cashier, who could potentially introduce error into the process--i.e. give you a Double Double Cheeseburger when all you wanted was a hamburger.
Panera Bread, Chili's and Applebee's are on board with concept, as they've rolled out--or plan to roll out--tablets that let customers place their orders and pay for their food without ever having to interact with a person.
Unfortunately humans are also prone to err in more serious situations--like deciding whether or not a patient is in need of open heart surgery. Khosla referenced a study that asked a group of doctors to review patient data in order to determine whether the individual was in need of cardiac surgery.
"They asked 40 cardiologists the same question about the same data. Half said yes, and half said no," Khosla said. "Whether you get heart surgery depends on what doctor you happen to pick. That's pretty bad."
But that's not the worst part, he said. "Two years later they took the same data to the same cardiologists, and 40 percent changed their minds."
Ideally machines will one day be able to consider thousands to millions of data points from one patient in order to definitively answer the question, "does this individual need heart surgery?"
Khosla indicated that all of this was to say that no one is immune from the seismic changes technology will have on the workforce. "Whether you're a farm worker to a warehouse worker to a sophisticated, well-educated medical practitioner--machines are quickly developing these capabilities," he said.
"I think it's likely over the next 40 to 50 years that more than 50 percent of all jobs that exist will get displaced," Khosla added.
Which is not something he takes lightly. If society thinks economic disparity is a problem now, he suggested that we haven't seen anything yet.
"We can't afford to let that happen. So something will have to change very dramatically." Khosla said he wouldn't go into any more detail because it was the wrong place and time for the discussion. But he added, "It won't be a technology solution. It will have to be a social solution."
Silicon Valley's risk appetite
Editor in Chief of MIT Technology Review Jason Pontin, who was moderating the discussion, asked Khosla if he thought Silicon Valley VCs were beginning to lose a true appetite for risk.
"I like to say that I don't mind failing, but if I'm going to be successful, it better be consequential," Khosla said.
Without naming names, Khosla suggested that too many venture capital firms invest in companies that look good on paper and are likely to succeed. And they may very well do that in the short term. But that doesn't mean they are worthy investments, as their long-term impact could be nil.
"Intolerance for failure drives you to the point where the probability of success goes up, but the consequences of success are mostly inconsequential. Think about that," Khosla said.