Most of the cool things that make our lives better have a downside. Most people like the convenience of texting but some do it when they should not -- and walk or drive in horrible ways. Examples of two-edged swords are easy to find. But most of us do like painless dentistry.
"If a driverless car runs a red light, who's to blame?" I have no idea and I hope judges and lawmakers can work it all out. Fixing blame and liability are huge. The annual cost of auto mishaps is in the hundreds of billions of dollars.
What about the Peltzman effect? Over a hundred studies are referenced in Google scholar. My quick survey shows the weight of the evidence is that the effect is real: those who know their brakes are more reliable (or that their seatbelts will protect them) will take more risks. It is found in NASCAR competition as well as in the use of visors on the ice rink.
If the odds that other cars are programmed to watch for (and automatically avoid) you go way up, what will you do? Will you revert to your bumper-cars style? Is there a tipping point?
At some point, if enough car owners in your town have installed lo-jack, there is no point in paying to have it installed. Thieves know how to play the odds and move on to another town (or trade). One can free ride. But with driverless cars, the bumper-car wannabees will surely avoid getting that new car because of the anticipated fun of being reckless but safe.