I see your point about the tone, but it's true that widespread adoption of such technology would yield interesting new technology for Google too. As a simple example, shifting traffic patterns are a useful leading indicator of economic change at a very high resolution. You can get a lot of statistically useful data from a surprisingly small sample size (http://en.wikipedia.org/wiki/Sample_size_determination). Consider, for example, the length of time cars spend in the parking lot at a shopping center. When you have that for hundreds of individual cars per day and/or longitudinally (over repeat visits) you can infer an awful lot of data about the local business climate in a non-intrusive way. Even if you make the cars wholly autonomous and don't collect data on individuals there's still a lot of environmental data that can be usefully harvested.
