A dark side to the “gig economy”?

Last January I wrote an article titled ‘An inevitable response to the robot revolution?‘  In it, I linked to a Telegraph report about the rise of the ‘gig economy’, where workers were self-employed in the new ‘sharing’ economy with companies such as Uber, AirBnB, etc.  I postulated that this might be one response to the automation of many lower-level jobs as robots and artificial intelligence became more widespread.

However, there’s another view of the situation – one much less encouraging to workers in the ‘gig economy’.  CNBC reports:

The so-called gig economy will cease to exist in 20 years, according to a new report from venture-backed start-up Thumbtack, an online marketplace that helps skilled workers find customers.

The study predicts that logistics companies — from start-ups like Uber to tech giants like Amazon — will soon replace drivers and delivery workers with autonomous vehicles and drones. Highly skilled workers, such as lawyers and accountants — no longer guaranteed jobs at big firms — will be the new gig economy workers, the study finds.

. . .

Uber is upfront about its plans to replace drivers with robots over time. “Autonomous driving technology has the potential to drastically reduce deaths in cars and make transportation even more affordable,” an Uber spokesperson told CNBC. “That’s an exciting future and one Uber intends to be part of, but that transition for technical, regulatory and adoption reasons, at scale, will take some time.”

. . .

In the meantime, the gig economy is creating invaluable data to feed Uber’s algorithms and build artificial intelligence systems — the brains of those robots. For example, an Uber driver is sending back a lot of data on where customers are as well as traffic and road conditions.

“All these things might ultimately enable the autonomous vehicles that Uber is very actively pursuing to better complete those kinds of tasks,” said Osborne.

“This gig economy — in that it is being pursued through digital platforms — is actually getting people to automate themselves out of a job through delivering data back to the platforms that can be used to provide an automated alternative,” he said.

There’s more at the link.

I’ve been warning for some time about the number of jobs likely to be automated in the near future.  It’s already happening, and is going to happen even faster as technology matures.  Unfortunately, the economic, social and cultural support networks to help those automated out of a job have not kept pace with the speed of events.  What’s going to happen to welfare and unemployment benefits if another few million workers are automated out of their jobs next year?  And more the next?  And even more the year after that? It’s a question we’d better be prepared to answer, before millions of unemployed ask it at the tops of their unhappy voices.

This also holds implications for the millions of people seeking refugee status in Europe, and the even more millions trying to sneak into the USA illegally. They’re all relying on the existence and availability of jobs to support themselves. What happens if the supply of low-end jobs is cut in half, or worse? Many such people won’t have registered for and/or be eligible for unemployment and welfare benefits. The possibility that they’ll turn to crime and violent protest is very real. Are we prepared to cope with that? More and more, illegal aliens and alleged ‘refugees’ (who all too often are economic migrants rather than truly refugees) are becoming a multi-faceted problem. I think those who advocate blocking them at the border, and evicting those already inside the borders, will find growing support among citizens and legal residents for their policies.



  1. Google's AlphaGo program just beat the current world champion Go player Lee Se-Dol in the first of the five matches. Go is a couple of order magnitude more difficult than chess which can be bruteforce compute. I'm actually very impressed with that performance.

  2. I grew up reading Asimov's robot stories but one thing always bothered me; with all these wonderful robots, what was there for people to do?

  3. Sorry Peter but I've worked on too many printers during my (ongoing) career as an IT admin to have any faith in seeing that kind of automation revolution in my lifetime.

  4. I'd agree that automation is coming for many fields.

    I'm not convinced that it will happen quite as fast as you fear. If nothing else, because for many of these low-skill jobs one of the main human contributions will be "respond to unexpected situations, and warn off human malefactors".

    I'm not at all happy about that last – the security guard economy? – but I suspect that jobs like "Uber driver" will evolve in that direction. Or dealing with accidents that leave a car incapacitated in the middle of a street, or a jammed automatic burger flipper, or ???

    Jobs that require judgement, training, or human interaction will remain human. Jobs that are both necessary and hard to automate will remain human. And jobs that can be automated will need a (smaller) number of humans to control or service the machines.

    I suppose I'm saying I agree with you on the result, but not the speed or completeness.

  5. for many of these low-skill jobs one of the main human contributions will be "respond to unexpected situations …"

    Yep, just look at those self-check stations in grocery stores, drug stores, Home Depot, etc. You still need a person to unravel the issues that crop. Granted, usually one person is supervising 4-6 stations, meaning there's a loss of 3-5 positions per store. Sometimes those people can be repurposed in the store, but most places that have made the switch to self-check stations don't have enough slack in the budget to absorb more than 1-2 people (if that many).

  6. Jobs that require judgement, training, or human interaction will remain human. Jobs that are both necessary and hard to automate will remain human.

    I'll agree with that, right until faster processors, less bloated and more task-specific software coupled with artifical intelligence solves the "need a human" problem, and it will, one function at a time.

    It may take quite a bit of effort, time and money to automate some human-specific tasks, but that hurdle has to be breached only once per task; from then on it becomes a economy-of-scale issue to implement across each instance of that task.

    Some tasks never will be non-human – critical surgery on humans, extremely complex interactive software, and the like – but those exist today, and will continue to exist, in rather rarified air; the great majority will continue to suffer along at the lower alttudes where automation will become more cost effective.

    Which raises the question of how one might automate an adequate response to millions of human-controlled torches and pitchforks…..

  7. Assuming this "doomsday" scenario of automation comes to fruition (almost a case of when, rather than if) then capitalism as an economic model will simply fail.

Leave a comment

Your email address will not be published. Required fields are marked *