In 2009, Air France suffered the deadliest crash in its history. The short story is that due to a fairly common sensor failure, the autopilot disengaged. Pilots then made a series of questionable decisions that ultimately led to the crash.
So, who’s at fault, was the question? Should it be blamed on the human pilots or the technology that failed? If it’s the technology’s fault, what do we do about that?
Something rather interesting happens when humans get excellent assistance, whether that assistance is human or technological: we rely more and more heavily on the assistance, and as a consequence our own performance suffers. Academics and industry experts have been observing this phenomenon in the airline industry for years now (if you’re interested, this Vanity Fair article is an excellent starting point for exploring the topic), and it seems that nobody’s 100% sure what to do about it. The quality of the automation improves with each generation of aircraft, and yet we still have multiple people in the cockpit for every flight, at least for now. That’s mostly because no matter how good the automation is, it’s not perfect. It can only respond to factors we’ve taught it, and on every flight there is a possibility, no matter how small, that something will happen that’s totally unforeseen – unforeseeable, as far as the technology is concerned. And it is at this point that the humans in the cockpit will be required to intervene. This combination of technology and human intelligence has led to a massive improvement in air travel safety. It is a good thing, but it’s not autonomous. It also makes training people for this kind of work colossally difficult, but that’s a topic for another day.
Elon Musk says Teslas will be self-driving by summer. According to TechCrunch, we’re “getting close” to sci-fi like the Cylons in Battlestar Galactica. This is misleading. It’s also unfair. This kind of talk implies that unless and until technology gets to a point where it’s totally autonomous, able to do what humans can do, it’s somehow inferior. When did we get it into our heads that the ultimate goal of technology was to emulate humans? What happened to technology helping us do what we do, better?
There is a narrow but very deep gap between assistance and replacement. Musk’s software (and that of other manufacturers who haven’t made the same claims) may be able to control the car in ‘normal’ circumstances, but you’ll still need to be alert because weird shit happens on the road – other cars are piloted by people, who are unpredictable and sometimes totally irrational. Siri and Google Voice will get better and better at parsing what you say and scouring the internet for whatever it is you’ve asked for, but they very likely still won’t understand if you live equally in two places (like I do). That’s because they’re only learning what we’re telling them to learn. And that’s OK. That’s great, even.
How often has it been said that good help is hard to find, that assistance is a huge proportion of success? That’s exactly what these technologies are, and they’re borderline miraculous. And yet we’re always looking forward to a time when all these things can think for themselves – when they will do for us (or to us, if you’re feeling dystopian), instead of helping us do. Funnily enough, it’s that point of technological evolution that some of the world’s great minds are urging us to think more carefully about, prepare more carefully for.
In the meantime, I hear a lot of poor design decisions (and their effects on people) waved away with a curt “yeah, it’s getting better. The AI is learning.” I’m sure it is, but does that mean that the experience of this technology that’s meant to assist me should be poor in the meantime?
When we get swept away by what’s next, we miss what’s happening today and what we could do to improve it. Yes, it’s important to see things in context, and some of that means thinking forward, but there’s a difference between considering the future and neglecting the present. If we’re to make technology that improves our day to day lives, we need to stay focussed on what it is now – assistive. And a little bit magic.