The hype surrounding autonomous vehicles shifted up a gear recently, first with claims that Google could develop its own autonomous taxis, and then with Nissan’s promise to sell an autonomous vehicle by 2020. While I hate to rain on anyone’s parade, autonomous or otherwise, I think these two stories should be taken with heavy dose of road salt.

The first story builds on a report in the German newspaper Aktuelle Nachirchten suggesting Google is working with auto-component companies, including Continental AG and Magna International. Google may well be considering all possibilities, but I think it’s quite unlikely it will attempt to build its own cars. That would be a colossal undertaking, even for such an ambitious company; and it’s experience with electronics hardware hardly seems like sufficient preparation for a leap into the immensely complex and high-risk world of automotive manufacturing.

It’s more likely, I think, that Google would develop some sort of self-driving control system in collaboration with a company like Continental, and that such a system would be offered to car manufacturers. Continental already supplies all manner of components and technologies to car makers, including various pre-crash sensors, and “driver assist” technologies for adaptive cruise control, blind spot detection, and lane departure warning. While a number of car makers are talking about their own autonomous driving technology, there are several that aren’t, and these laggards could well be interested in some sort off-the-shelf solution.

Nissan is one car maker that clearly isn’t shy about talking about its plans. But I think recent coverage concerning the company’s autonomous driving goals is confused about what the company is actually promising. Its statement says that it will introduce something called Autonomous Drive by 2020. This is unlikely to involve full automation—i.e. total control of steering and braking—100 percent of the time. It’s more likely to mean full automation in limited situations, particularly highway driving, and this will probably still need to be monitored by the driver while in operation. No-one I’ve spoken seems to believe it’s yet possible to make a car capable of constant, full automation. If Nissan were actually able to deliver that, then I’d say the hype may well be justified.