For the past decade or so, we've been warned that driverless cars will change the world of transportation. This is all egregious nonsense. All that's been achieved, or is likely to be achieved any time soon, is something more akin to enhanced cruise control rather than anything resembling true driverlessness.
Today's "autopilot" (a term that only Tesla uses) is really only practical in places where traffic is extremely predictable such as highways. Even then, the driver is supposed to remain fully aware of what's going on and take control if something goes wrong, because as Jason Levine, executive director of the Center for Auto Safety, recently pointed out, "these advanced cruise-control features Tesla has are not very good at detecting and then stopping for a vehicle that is stopped in a highway circumstance."
Every automaker except Tesla uses GPS to prevent drivers from using "driverless" features on city streets and, as the New York Times recently pointed out: "Tesla owners' manuals warn customers not to use Autopilot on city streets [because] 'failure to follow these instructions could cause damage, serious injury or death.' "
The fully autonomous car envisioned as the "transportation of the future" is sometimes called "Level 5"; the most that's been achieved (or is likely to be achieved) is "Level 2," which is essentially nothing more than advanced cruise control. We are nowhere near Level 5, because true driverless cars require A.I. that has common sense.
Look, I've been following and writing about A.I. since the early 1980s, and there have been no breakthroughs in "hard A.I." as it used to be called. Seriously, the kind of object avoidance that's available in Level 2 of "driverless cars" was available in toy robot kits in the 1990s. I know because I used to build them.
Folks, there haven't been any huge breakthroughs in A.I. The "singularity" is not near; it's as far away as it's always been. What has happened are incremental improvements in image recognition, aided by the ability to draw on large data sources, along with "self-learning" algorithms that work inside closed systems with easily identifiable rules.
Social-media moderation with A.I., for example, is smoke and mirrors; the real work is done by human beings. Customer service A.I. is horrible and mostly useless. Voice recognition remains a joke. "Autocorrect" is proverbially stupid. The "singularity" is no nearer than it was 50 years ago, when it was predicted first predicted to be coming "in 10 years." Since then, A.I. with common sense perpetually remains "10 years" away.
Here's the real question? Why does the A.I. myth (in its "driverless car" manifestation and many others) continue to remain popular in the face of overwhelming and constant evidence that it's just not happening?
I have a theory.
Throughout history, men (and I do think this is almost exclusively male) have been obsessed with the idea of being able to create human life without relying upon human females to create it. You see the concept in legends of talking statues in the Greco-Roman period, in the alchemical legends of the homunculus, and of course in science fiction.
Science fiction examples of A.I. are almost always either sexbots (from Fritz Lang's Metropolis to Spike Jonze's Her) or sexless eunuchs (from C3PO in Star Wars to Robin Williams's Millennial Man). Curious, isn't it, that there is almost an entire lack of A.I. characters in science fiction who have functioning male genitals?
Maybe when it comes to thinking about the future of "driverless" cares we should take aggressively macho uber-dudes like Elon Musk (and his thousands of imitators) out of the decision making loop on the subject of whether "driverless cars" are ready for prime time.
Let's face it, testosterone and driver safety have never been a great match.