Are self-driving cars just around the corner, or years away? The media and certain business leaders are hyping the technology in the pure optimism camp, while academics and other researchers seem much less sure.

On the optimism side, you can now summon a Tesla to your front door from its parking spot, or send it from your front door to go park, with the “Autopark” feature.  And Elon Musk calls this just a “baby step” with more ambitious self-driving features just around the corner:

Tesla Motors Inc. Chief Executive Officer Elon Musk said within two years a customer would be able to summon an electric car to drive autonomously from Los Angeles to New York.

“I might be slightly optimistic on that, but I don’t think significantly optimistic that we can do that in two years,” Musk said in a press conference on Sunday. He added that it should be technically feasible to have fully autonomous vehicles within 24 to 36 months.

But then you read reports of self-driving researchers who throw cold water on this kind of optimism:

Everyone wants to reduce the global death toll of 1.2 million from car accidents annually, but how do we get on the path to vehicles that are virtually incapable of crashing? There’s a lot of cheerleading out there, and vague talk about vehicles being on the road in a year or two. “In truth, we’re a long way from autonomous cars,” [Dr. Gill Pratt, head of the new Toyota Research Institute] said. “We can perform reliably at certain speeds, and under certain weather conditions. But what we’ve accomplished are the relatively easy things.”

Pratt set the industry a rather daunting goal. Sure it’s great that autonomous cars have been tested in millions of miles of driving, but we need trillions of miles on the road. Figuring out how to make that happen is the team’s challenge. Part of that is measuring how humans react in emergency conditions while piloting the cars already on the road. Pratt pointed out that Toyota, now the world’s biggest automaker, currently sells 10 million cars around the world annually, and that 100 million Toyotas are on the road at any one time—covering a billion miles per year.

Hard to know what to make of it, but it could be a case where business leaders like Musk pull inherently cautious researchers like Pratt out of their comfort zones to bring the technology to market faster than they’d otherwise like.  The result could be that we end up somewhere in the middle of these predictions.