Google's modus operandi is not so much to sell people products as it is to provide them services. What if Google cars weren't owned or leased but provided a distributed driving service like Uber?

Couldn't the terms of that driving service stipulate that it would act in a certain way which might not prioritize preventing an accident to any particular car but would minimize injuries/deaths as much as possible for all users of the service? I'm not familiar with liability law like you seem to be so I'm honestly asking.

Alternatively, maybe Google could sell you the car but you would have to choose which automatic driving data service to connect it to. If you choose not to connect it to any automatic driving data service, it would act like a standard human-controlled car. But if you chose to connect it to a driving data service, that would come with a license agreement that the data service would control your car in a way that could injure or kill you.

I think it's a simplification to think of googlism as an innate aspect of the car/product when it's more a dynamic interaction between the car and an expansive data service. Even their current cars need a constant connection to work.