The technology, while not perfect, is already here: "sniffing" sensors can detect trace amounts of alcohol in an enclosed environment; skin sensors -- mounted on a steering wheel or an armrest -- can detect the same on the epidermis; driving pattern and facial recognition looks at vehicle operation patterns and facial movements to determine impaired function. Soon, some or all of these technologies could be part of your daily driving experience.
Of course, few would argue that less drunken drivers on the road is a bad thing, but the technology, or rather the possibility that it will be a required part of your vehicle's cabin, raising some questions about a person's right to privacy versus the greater good -- not to mention technical questions about how the tech operates in real-world situations. After all, drunken driving is already illegal, as is underage driving. If you had to scan your valid driver's license before operating a vehicle, do you think people would take that, too, as an invasion of privacy?
As for the drunken driving issue, it's also what happens with the information -- would it be stored in the vehicle's computer? If so, for how long? Would your insurance company have access to the fact that, while you were unsuccessful, you actually tried to operate your vehicle while under the influence?
As for the technology itself, how does it work in real life? The sniffer- and skin-sensors (and possibly the facial recognition application) could conceivably stop you from starting your car if you're already inebriated, but the driving pattern recognition requires you to already be moving for it to determine if you're sloshed. And what about sober drivers who start drinking on the road? Does the car shut down the next time you come to a stop? What if you're on a long stretch of highway? Would it alert the authorities?
As you can see, in the case of in-car alcohol detection systems, creating the technology is probably the easy part -- the hard part comes with the questions raised by the endeavor.