Slashdot Mirror
Self-Driving Cars Could Cost America's Professional Drivers Up To 25,000 Jobs a Month (cnbc.com)
The full impact of self-driving cars on society is several decades away -- but when it hits, the job losses will be substantial for American truck drivers, according to a new report from Goldman Sachs. From a report: When autonomous vehicle saturation peaks, U.S. drivers could see job losses at a rate of 25,000 a month, or 300,000 a year, according to a report from Goldman Sachs Economics Research. Truck drivers, more so than bus or taxi drivers, will see the bulk of that job loss, according to the report. That makes sense, given today's employment: In 2014, there were 4 million driver jobs in the U.S., 3.1 million of which were truck drivers, Goldman said. That represents 2 percent of total employment.
On the other hand, if they can retrofit their own rig to be self-driving, wouldn't that turn their rig into a money-maker?
#DeleteFacebook
One of my Civil Engineering prof's told us that one truck does the damage of 10,000 cars
I second this. In south Texas a recently discovered shale formation (the Eagle Ford Shale formation) created an oil boom. This caused tons of oil-carrying trucks to just completely ruin southern portions of highway 183 to the point where it's damn near unsafe to even go the speed limit anymore.
It's harder to fly a plane or helicopter. There are a lot of things to hit on the ground, a lot of information feeds, and a lot of decisions to make; the air is nice and clear, except for invisible turbulence, stalls, and other situations requiring massive processing of information in ways not well-documented.
To get planes to self-fly reliably, you have to make them not drop out of the sky in a stall. Pilots do that by experience, which is just knowledge and an interpretation of feedback. Since we don't have a way to explain the generalized algorithm and information set pilots use, we could, at best, use complex flight recorders and bayesian analysis to generate statistical models which attempt to use only the specific situations encountered plus a limited degree of extrapolation on variables we've identified as relevant. None of the indicators are visual; we can only pull values from temperature sensors (which are slow to react to temperature changes), accelerometers, gyroscopes, pressure sensors, and stress sensors (i.e. power meters) attached to the movable parts of the plane to work out the situation. That means we have to either hope for a simple correlation between these variables or find a transformation algorithm to match them to what the pilot senses.
Cars can sense wheel rotation speed and identify when individual wheels are slipping. Accelerometers, gyroscopes, cameras, LiDAR, and prescriptive data feeds (e.g. maps, GPS) give you a pretty good sense of how the car is moving. You can tack on things like stress sensors on suspension components to model vehicular forces, and current models don't even do that--it might not even be necessary. Vehicle dynamics are pretty easy to work out from the way the car is moving now and the amount of wheel slippage; aerodynamics are negligible, so invisible forces aren't going to send your car spinning out of control or cause it to slide along the road due to a loss of traction.
As for the replacement rate, 25,000 per month isn't a lot. There are 192,000 freight trucks sold per year, or 16,000 per month. That leaves 9,000 taxi cabs or other such things.
It's not a big deal at that rate, anyway. The job turn-over is actually pretty high, and this gives a lot of recovery time. It's only 0.0166% of the workforce per month, and the adjustment rate for new contracts to push down shipping costs should pick up as soon as someone can scratch into a market--which means a freight company could even start expanding to weaken a competitor by deploying more trucks than the drivers it's eliminating and cutting its shipping pricing to attract more business. The added volume, even with the margins the same, will grow that company's cash flows and make them more capable of taking actions to gain market traction--while the competitors will have to lay off workers who they don't replace with self-driving cars.
In other words: we should see some job replacement in 2-3 months due to a slight reduction in shipping costs putting a control on consumer prices (i.e. prices rise slower than consumer wages; they'll slow their rise just a tiny bit more), but it's not going to stop the growth of unemployment at that level. It could be 6-12 months before the competition in the market really starts driving prices down, and those input costs start leading downstream businesses to price competition. We may see a full swing of 0.1%-0.2% unemployment at peak with a transition rate of 25,000.
Once that replacement rate kicks in, the rate of transition onto autonomous cars will pick up as a market imperative. It's two-fold: slightly-lower costs mean consumers can buy slightly-more, and part of that goes into increased shipping demand, which means labor on operational support, mechanics, fuel (electricity), and so forth. In total, it's still less labor in shipping, and less labor per unit shipped. Anything shipped must be sold (retail), as well, so some of the labor goes there. Even then, you've got slac
Support my political activism on Patreon.
As someone who used to drive big rigs for my dad back in college, I can say that anyone who thinks an AI will be able to drive a modern tractor-trailer anytime soon has obviously never driven one. A tractor-trailer is about 100 times more difficult, complicated, and dangerous to drive than a regular car. And we don't even have AI's that can reliably drive cars yet. Shit, they've only just recently developed reliable automatic transmissions for those beasts.
You just show me a AI that can safely and consistently alley-dock a 62-ft trailer down some ancient one-lane road with a turn-in that the trailer can barely even clear, in a city filled with unpredictable traffic and 4-wheel drivers who HATE waiting on tractor-trailers and don't care about traffic laws.
SJW: Someone who has run out of real oppression, and has to fake it.
I don't think anyone thinks they will dock themselves soon, but it would still be a great money saver if they could park themselves at a hub and wait for a couple jockies to drive a foldable scooter out and bring them in. Eventually docks may be designed to accommodate self driven trucks rather than the other way around. I could see Amazon doing something like that.
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
I could also see Amazon partnering with one of Musk's companies to build hyperloop for freight. It seems like building a 1-meter or even 30-cm freight pipe would be a heck of a lot easier than transporting people. 1-meter could fit almost everything they sell, and 30cm would still be useful for a lot of products. We'd get an operational test of the hyperloop concept. The train people would really sweat bullets over that one; but I'm not sure where they'd acquire the rights-of-way.
For all intensive purposes, "whom" is no longer a word. That begs the question, "who cares"?
I third this. Some studies of highways built to Interstate Highway standards in the US show cars can't wear out the main road and would take over a century to wear out bridges and other connectors. Trucks are what break roads. And in this amazing fact based world, we subsidize long haul trucking, and we are systematically dismantling over a century of right-of-way building destroying critical infrastructure. Even the military mothballs vehicles just in case... In the Seattle area, by way of example, on the east side of Lake Washington, the right-of-way for rails owned by various regional agencies have systematically, deliberately, and with no thought to the future dismantled the railbed that could be used by the current light rail effort. This will increase the cost of light rail dramatically. (not to mention that light rail is not a good economic fit for the region, they spend billions and more billions of a project doomed to eternal subsidy.) The rail system could also act as a "backup" for the tracks through Seattle to allow for needed reformation there.
No planning just money grubbing and empire building. Can't very well build an empire if someone else owns the infrastructure!
- Tjp
I am in wallow with my inner money grubbing capitalistic pig. ... Oink!
Exactly. Automating trucking (and other transportation) would be a huge boon to our economy, not a drag on it. Suppose for a moment that in one day, every truck was capable of moving itself around automatically, sans person. What do you think will happen to the cost of shipping goods? What will happen to the volume of goods moved? What does that do to the volume produced / consumed? There may be 3,000,000 truckers, but there are 300,000,000 consumers, and everyone of them benefits.
These stories are very one sided and usually portray the losing side. Just like crying for the buggy whip manufacturers when buggies got petrol-powered engines.
Food will cost less. More people can therefore afford to eat. This is a good thing.
I'm sure a lot of criminals who don't have the gall to assault a regular truck may be able to justify going after a self-driving truck, since there are no people onboard to leave behind as witnesses.
Well there's also nobody to intimidate. Nobody with any keys or codes to give you access to or control over the truck. My first thoughts apart from the constant cell phone/GPS tracking to alert police would be to just kill the engine, lock the brakes, give a little light and siren show and if you can't draw anyone's attention and they're really determined to break in by force before the police get there, just set off a few dye packs/stink bombs. Sure it'll ruin the cargo but zero payoff will make the highway robberies stop pretty quick.
Live today, because you never know what tomorrow brings
> cannot be AT LEAST as flawless and safe as a human vehicle operator, then it has no business operating a vehicle at all.
Leaves lots of room for computers. Doesn't have to beat the best driver in their best condition, just has to beat the average driver, the sleepy drugged up ones, the vindictive ones...
Their are many things autonomy beats your average driver at today, and getting that on the road will be a big advantage. That is obviously step one (same as light vehicles today) Get it to save sleepy drivers from leaving their lanes, get it to slow down rigs driven past their safe limits, get it to warn of hazardous drivers and conditions... Then it will be take over the low hanging jobs like clearing railway and shipping terminals. Take over long haul interstate, so one driver can can handle more miles safely. Like oil and coal power, Trucking is likely not sustainable (at least at current levels.) So it does need re worked anyway, so they will figure out what can be automated and what can be optimized, and eliminated that a computer may not be able to handle as well. Since computers can control many more variables more precisely likely that will result in trucks and docks, and containers optimized for those conditions, and removing those hard for automation.
Autonomy is already controlling bigger rigs with more precision than 90% of truck drivers today can. (3500HP mining trucks going over 50 mph with a million pounds carrying thousands of gallons maintaining 3" precision in backing, also railways, steel mills, ships.) So yes the software and hardware is not complete today for OTR, but 20 years ago most people said internet banking would never happen also.
Horse shit. The average cost of moving freight via truck averages between $1.60 - $2.10 per mile. An excellent driver with ten plus years experience will make maybe .45 per mile. The majority of drivers make less than 35.
I work in a specialized part of this industry where an average move is $5K. Of that the driver makes around $1K.
Do the math. ‘Trickle-Down-Economics’ has never worked in the real world. The consumer will never see that cost savings. Marketing bullshit from companies that pretend to care about you, notwithstanding.
Your sig here!