Halo: the SpinVox of autonomous driving

Last week I was in Seattle, and the good folks from Halo were demonstrating their rather neat tele-operated car sharing system. This is not autonomous vehicles, although it might look very much like it. This is remotely piloted vehicles that are delivered to the driver, who then drives themselves to their destination, at which point they abandon the vehicle to be remotely piloted to the next job. As someone who, in the last few weeks, has spent a lot of time picking up cars from airport car rental locations, this actually seems rather appealing as an idea.

What it certainly reminded me of was the voice-to-text translation company called SpinVox, whose light shone brightly in the mid-00s. If you’re not familiar, that company provided a service whereby received voicemails were rendered into text and forwarded on to you as an SMS. I loved this service, incidentally, despite the fact that it was far from perfect. The controversial thing was that it transpired that rather than it being a fully automated speech recognition technology, actually there were banks of workers in call centres listening to the calls and typing in the text. There is some contention about the split between what was automated versus what was done manually, but the company never quite recovered from this revelation and its attempts to obfuscate the way the translation happened.

Don’t get me wrong though, Halo is transparent about how it operates. However, it’s another example of providing a service that should be deliverable using sophisticated technology, but actually still also relying on using people (often in call centres). In the case of Halo, it’s more efficient than taxi drivers, because they’re only engaged in the act of getting the vehicle to the passenger and picking it up again afterwards. And, it should be noted, Halo says that all its drivers have 6+ years’ experience of driving in the location in question. That, plus the latency limitations mean that this is unlikely to be outsourced halfway around the world.

Of course, there are some practical questions still to be answered. How much redundancy do you need in terms of number of drivers in order to be sure to be able to cope with peak demand? Halo has to provision for the point of peak demand, the ‘busy hour’ (as us telecoms folks would describe it), meaning it may have lots of spare capacity at other times. And also, surely that this is an intermediate technology. But that’s fine if Halo wants to pivot in the future it will be able to do so.

There tends to be an assumption that there is a bright autonomous future ahead of us. But even in relatively controlled environments, such as railways, there seems to be a necessity for human involvement. We’ve seen numerous examples of robo-chefs too, but how many of us really want our dinner cooked that way? The darker side of this, and not something of which I’m accusing Halo, is that there are a lot of technology evolutions that succeed principally because they tap into paying lower wages. Uber, for instance, is a success mostly because it’s cheaper than taking a taxi. And it’s cheaper because the drivers mostly earn less than traditional taxi drivers. And also, because Uber makes a big loss every year. It is, of course, a neat app that’s very convenient to use. But let’s not avoid the fact that a lot of its success is down to being prepared to squeeze the driver and itself in order to keep prices lower than its competition.