Jobs that are safe from AI

Will AI replace Van Drivers?

The question of whether AI might replace the job of a van driver is more complex than evaluating some other jobs on this site. On the one hand we're all aware of the many efforts to develop fully autonomous vehicles, but on the other, those efforts are still a long way off - at least for the kind of autonomy needed for this particular role.

There's other technological advancements that play into the question - aerial delivery (by drones), sidewalk-type delivery robots, and humanoid robots that could potentially handle the loading and unloading aspects, or even handle actual deliveries themselves.

This article considers all of these factors, and the many other skills that go into being a van driver, before arriving at its conclusion.

Short version... the job of a van driver is safe from AI for the foreseeable future.

What kind of van drivers are we discussing?

If you have a job as a van driver it means you are most likely involved in deliveries. There are of course many small business owners who own vans to operate their business, and many employees and contractors who drive vans for the purpose of transporting tools or specialized equipment to job sites. Such van drivers are not really considered, as the driving of the van in this case is secondary to the primary job.

The practical aspects of being a van driver

To consider the potential threat to van drivers by AI, it's helpful to go through the practical aspects of the job, as any AI would need to do most (if not all) of these things.

Checking the van is safe and ready for a day's work
Basic maintenance tasks
Loading the van, either at a depot or via pickups
Communicating with other workers or members of the public at pickup
Dealing with paperwork and admin
Checking packages and that others haven't made mistaks
Loading the van
Organizing the van
Driving the van (a diverse range of routes and conditions)
Finding the right delivery location
Dealing with unexpected issues at the delivery location (barking dog, angry recipient, blocked access)
Making deliveries at commercial and/or residential locations
Communicating with people and members of the public at delivery locations
Finding the right package or goods for a particular delivery location, within the van
Unloading the van
Carrying packages and other goods into the delivery location (dealing with an endlessly diverse range of situations, loading docks, and sometimes going into buildings)
Taking photos of packages at the drop off locations (proof of delivery)
Getting signatures from recipients
Fueling the van (gas/diesel/electric)
Securing the van and its contents
Dealing with unexpected issues en-route (e.g. No one in, wrong details, flat tire, traffic wardens)

This is a long list of some fairly diverse things, that should give pause to even the most bullish proponents of AI technology. It's not just the problem of autonomous driving that needs solving here.

The Autonomous Driving Challenge

Even if it was just the problem of autonomous driving that needed solving, this is an immense challenge that has humbled the smartest eggheads in Silicon Valley. In the mid-to-late 2010's for example, there was a ton of hype about fully autonomous driving being 'just around the corner', but for anyone with half a brain who'd actually driven a vehicle in their lifetime, it was clear that the idea of 'go anywhere', deal with any situation, fully autonomous driving, was a quarter of a decade away, or more.

That's because while any idiot can drive a car, the act of driving uses an immense array of distinctly human senses, experience and intuition, that a computer has an incredibly hard time matching.

How for example could a computer handle the following driving situations?

You spot a pedestrian approaching an intersection who is staring down at their phone, not looking where they are going
You see two kids playing in a field up ahead with a ball
You catch a glimpse of movement behind a parked car, and life experience tells you an animal might be about to jump out
Two teenagers are staring down at cars suspiciously from a bridge ahead, with what looks like something in their hands
A truck coming the other direction loses a wheel and it's bouncing towards you
A plane having an emergency lands on the road in front of you
You spot a pothole in the road ahead that, is filled with rainwater (computers can't see this)
You spot downed wires across the road ahead (as above)
You turn a bend on a country lane and come across a herd of sheep
All manner of temporary roadworks, diversions, contraflows, temporary traffic lights
Police officers giving manual traffic control directions that contradict traffic signals
Flash flooding on roadways
Boulders across the road due to flash flooding
Dealing with a parade or protest that's filling the street
Dealing with aggressive drivers, drunk drivers, those weaving in and out of traffic
Extreme weather conditions - thick fog, icy roads, tornadoes
Maneuvering around broken-down vehicles, obstacles or unusual accidents on the road
Reacting to hand signals from other drivers
Dealing with malfunctioning traffic signals
Reacting to emergency vehicles coming up behind you, who need to get by fast
Maneuvering through toll booths on highways that require manual payment
Situations where GPS signals are lost, or the maps are outdated
Interacting with people directing traffic at construction sites and special events
Dealing with vehicles that are reversing
Dealing with blocked tiny streets and alleyways (even more so in Europe)
Recognizing and reacting to road signs that have been vandalized, obscured by foliage, or knocked down by accidents or weather

As a technologist and 40 year software developer who has dealt with every one of these situations while driving (ok perhaps not the plane landing, or Tornado), these are immense technology challenges to solve.

Frankly it will be far easier to solve fully autonomous flying cars, long before we solve fully autonomous road driving.

In my opinion we are at least 20 years away from solving this immense challenge, even with the recent mindblowing advancements in AI. Consider how long it took to solve voice recognition - the first mainstream voice recognition software came out in 1990 (Dragon Dictate), and this basic human skill has still not been replicated perfectly by computers, 30 years later.

Humanoid Robots

Fully humanoid-type robots powered by AI are coming, and like something out of a science fiction movie, we will literally see them on our streets around 2030-2035. It's a bit shocking. A ton of jobs are going to be disrupted by this and AI, hence this website, and it's not hard to imagine that loading and unloading robots for example, will be on the market in that timeframe.

They'll be very expensive at first, and not very adaptable, but that will change eventually.

But looking back at the first set of bullet points above, it's still hard to imagine a robot or other form of AI, being able to handle all of the skills needed to do the job of a van driver anytime soon. A few aspects perhaps, but not all of them. And it kind of needs to do all of them to replace the van driver's role.

What is more likely is that we see humanoid robots on our sidewalks making deliveries, or other robot form factors like the Starship delivery robots, which also use sidewalks, and are already in use in places like San Francisco. But these robots are slow and can't handle more than the smallest of packages. So not really a threat to the average van driver.

Delivery by Drone

This is an area that will have some impact, as the technology to deliver packages by air is already with us, and making steady progress. Whether regulators will allow widespread use of it in lots of places remains to be seen, and whether the public will put up with the noise of 'swarming bees' everywhere is another question, but technology-wise it is arguably a more likely near term possibility than the other technologies above.

An interesting approach to drone-based delivery was recently demonstrated which lowers a package to the ground from a hovering drone above, and this solves a lot of problems around finding a suitable place to land, in a diverse range of situations, although it still doesn't deal with the problem of an angry dog attacking it.

Conclusion

If you are a van driver or run a local delivery van business, you really should not worry about AI affecting things, at least in the foreseeable future.

It's of course always a smart idea to stay abreast of the latest technologies, as there may be areas where AI could actually help your business, and perhaps give you a leg up on competitors.

Using AI for route-planning, scheduling, marketing, social media, drafting business proposals, emails, helping with SEO, and other administrative functions, are all near term applications that you can benefit from today, or very soon.

Conclusion: Safe | Time Period: 20-25 years

Van Driver