WC Archive

There are numerous reasons for thinking that computer-augmented cars are more likely than computer-controlled ones. It's simpler to give computers narrow tasks than to teach them the gestalt of driving. Human drivers are probably less likely to resist the introduction of technologies to keep them safe in emergencies or to make annoying problems simple than accept relinquishing control entirely. The psychology and symbolism of having a machine assistant differs considerably from the psychology and symbolism of having the machine take over. And ultimately, there are some tasks at which people excel and computers don't, and vice-versa. Human reaction time is terrible compared to computers, for example, but humans are (for now, and probably for awhile) better at taking in the "big picture" of a situation and assessing possible dangers.

Most importantly, computer-augmentation is something which can be implemented today, not in a possible "once we figure out the bugs" future. We've discussed "adaptive cruise control" already: it adds a radar distance sensor to the cruise control system, allowing the vehicle to slow down if the car in front slows, and keeping the car at a safe distance. It's available across car companies, mostly in high-end vehicles. (A future version, "cooperative adaptive cruise control," would allow vehicle systems to communicate with each other, sharing road condition and driving information.) As another example, the Japanese version of the Prius has since 2003 had an "Intelligent Parking Assist" feature able to handle parallel parking (something many humans never quite master). Technology Review offers yet another, more recent, example: Nissan is considering the introduction of a "swerve avoidance" feature which would nudge the car back into its lane if it drifts over a lane marker. (A minor side-benefit of the swerve avoidance technology: it automatically disables itself if the driver puts on the turn signal. A technology which encourages drivers to use their signals is no small victory.)

Of these, the Nissan swerve avoidance may be initially the least popular, as it's the only one which takes control of the vehicle away from the driver (albeit only under certain, very limited, conditions). It's also the one most likely to get rave reviews, from people who have avoided accidents due to unintended vehicle drift. The question, then, is (if Nissan actually introduces this) will the voices of those who are saved by the technology be heard over the voices of those who are annoyed by it?

Computer augmentation is most welcome when it provides additional capabilities when requested. We're less happy when the computer assistant distracts us with its presence, and certainly unhappy with "nag" systems that ring bells when we disobey arbitrary rules. Augmentation that feels like Big Brother on Board will be just as unwelcome as augmentation that feels like the infamous Microsoft "Clippy" ("You appear to be driving to Duluth! Would you like some help with that?").

But here's the fundamental question: if computer augmentation has the capability to protect us from disaster, are we willing to accept a temporary loss of control? Or to put it another way, at what point does assistance augmentation become too much help? An under-the-hood radar system that automatically brakes if you get too close to the car in front of you would never make it to market; one which simply pinged under the same circumstances would be annoying. But a system which stayed out of sight until it calculated that at the current speed and distance a human driver would not be able to stop before an impact, and hit the brakes on its own, might be useful. But what about a less deadly, more common example, the idea of phones with breathalyzers able to prevent late-night drunk dialing? Or email software which pops up a warning if you try to send something that might get you into trouble.

Designers of systems for human augmentation should realize that what they're making takes away some measure of control even while it provides new capabilities. For many of us, it won't be an easy trade-off.