Which headline do you think would sell more papers?
INTELLIGENT ROBOTS KILL 20,000, NO SIGN OF STOPPING
AUTO FATALITIES DECLINE 50%
In a nutshell, this is the PR problem of technology. Technological progress is taken for granted, and technological problems are trumpeted to the skies. In this case, I’m pessimistic about the chances of autonomous cars. Here’s an article for Motor Trend which isn’t really about the safety of robot cars, but which takes the following shot anyway:
“Driving will be safer,” say the experts. “Computers will ensure that smart cars always maintain a safe distance between each other.” Uh, just a moment. In the shockingly brief amount of time I’ve spent writing this column, my PC has already crashed twice. Now just imagine 3000 morning commuters in their state-of-the-art, four-wheeled Intels, barreling down the freeway at 70 mph in perfect, computerized formation, when suddenly, for no apparent reason, the lead car’s robot coughs up the Blue Screen of Death. Personally, I don’t want to be at the site of the steaming wreckage when Bill Gates arrives to say, “Perhaps you folks would like to try our new SmartCar 2020 upgrade?”
Uh, just a moment. Robot cars don’t need to be perfectly safe. If 20,000 people a year die from computer crashes causing real crashes, that’s still about 20,000 or so fewer fatalities than us meat computers manage to rack up every year. Is the emotional response of the public going to agree? Like the headlines above, I doubt it. And this is true despite the fact that unlike people, purpose-built AI gets continuously better and better. The next year will have fewer robot car fatalities and the trend would continue.
This kind of sensationalist thinking has helped hold back the nuclear power industry. Despite the fact that coal power kills coal miners, hurts the environment, and releases radioactivity, it’s nuclear that people are afraid of. And that because of one failure, built by incompetent communists following a horrible design and then deliberately disabling safety mechanisms to run tests. Even then the total human loss was a few dozen killed in the immediate aftermath, and a few thousand cancer deaths in the years following. Nothing compared to the bloodbath on the roads, and in practice impossible with today’s technology. But here we are, with many people still terrified of nuclear power. Here we are again with the LHC, with all the news about the “end of the world”, and now the hysteria over a broken and fixable helium blowout. Technology needs better PR.
Here would be my strategy for the robot cars. Tie the robots into a GPS system, and have them only function on the comparatively easy interstate system. Once the public is more confident in the system, gradually extend the law to let the robots drive on stat highways, and then major city roads, and then finally everywhere. Over a few years I think the large numbers of saved lives would help avoid the dramatic “ROBOTS KILL” headlines. The point is to do the best job possible of avoiding dramatically bad headlines. Be gradual.
In conclusion, I want my robot car. It’s 2008. Let’s get going!