How a Forrester analyst thinks about the robots taking your job

robotsRobots and other forms of automation are going to transform the job market, but how much? There are four points of view. And that’s a good way to understand what it means to be an analyst.

  • The pessimist looks at the decline of jobs like secretary or factory worker and sees the same pattern in automation today. The pessimists write articles like like  HBR’s “What Happens to Society When Robots Replace Workers?” It’s a good way to get headlines.
  • The optimist believes that automation creates new jobs, jobs like mobile app developer that didn’t exist ten years ago. “Technology always creates opportunity,” the optimist says. Optimism is a nice quality in your friends, but is dangerous for serious business planners.
  • The realist recognizes that both job loss and job creation will happen. This is better than the black-and-white picture, but “it depends” is not a useful answer, either.
  • The analyst starts by asking questions. How many jobs? Which jobs? How soon? Answering these questions takes research, and leads to insights you can’t get by taking one side or the other.

Let’s look at how Forrester Research analyst J.P. Gownder tried to answer these questions in his new report called “The Future Of Jobs, 2025: Working Side By Side With Robots.” (Disclaimers here: while I used to work with J.P. at Forrester and actually worked with him on the very early stages of this report, this report is completely the work of J.P. and his Forrester colleagues. The report is behind a paywall, but J.P. also wrote a blog post about it, which is free. )

How do robots take jobs?

Look at “robots” as broadly as possible — include all types of automation, physical and virtual. It’s not just industrial robots that are changing the workforce. You interact with a kiosk in the airport instead of a gate agent, robots pick packages in warehouses, and IBM’s Watson makes cancer diagnoses. J.P. identifies three kinds of tasks that bots, both physical and virtual, can do, or at least facilitate: intellectual tasks (say, evaluating baseball statistics), customer service tasks (like self-checkout), and physical tasks (welding together parts of a car). Now we’re getting somewhere.

robots jobs

How many and which jobs are at risk?

To answer this question, you need to know what jobs people have now. The Bureau of Labor Statistics neatly tabulates and publishes this analysis every year. Start with that, and you can analyze all the major job categories and get an estimate that’s far more interesting than the guesses of the the optimist, pessimist, and the realist .

Forrester rolls these estimates up and gets a numerical answer. From the report:

[A]utomation will replace some jobs and create others. While automation will lead to a net loss of 9.1 million US jobs by 2025, that’s nowhere near the 69 million that many pundits have predicted.

What can we learn from the projection?

Don’t trust this number. Projections ten years out are unlikely to hit the target. There is just too much uncertainty. (Ever track the accuracy of government budget projections?) I’m not faulting J.P. — no one but Carnac the Magnificent can accurately project ten years out.

So, is this projection worthless?

While the number helps Wired to write an article and gets headlines, here’s how you should think about it:

A projection is story about the economy.

The value is in the details behind the projection.

For example, Forrester says automation will eliminate 22.7 million jobs, especially in construction, office support, and sales by 2025.

The report also suggests that bots will create 13.6 million jobs, or 9% of the workforce, in categories like software, engineering, support, or training. (Subtract that from the 22.7 million and you’ve got the net job loss projection.)

There’s quite a bit more detail than that, but I don’t have space to go into it here.

But now we’re smarter. If you’re running a college, designing a tax plan, arming a workforce with technology tools, or projecting unemployment rates, you need to plan for these changes, category by category. That’s a valuable tool that you can use in your planning.

How will robots change work?

Forrester’s report changes the conversation about robots. While they’ll cause both job losses and job gains, J.P. focuses on how they will transform jobs as people work with the new automation tools. Forrester thinks 25% of the jobs in every job category will be completely different as a result of automation.

Consider how the jobs of stockbroker, human resource staffer, and customer service representative now look completely different from the way they did 20 years ago. How will future jobs look different?

Retail looks pretty different with robots stocking shelves and salespeople armed with tablets as they help customers.

Finance looks pretty different when the systems run themselves and cough up insights that the financial analysts can interpret.

Health care looks pretty different when your doctor has a virtual assistant to make suggestions and take care of insurance rules.

Once you ask the question “How will my job look different once intelligence is available on demand in new physical devices,” you’re on the right track.

What’s the right way to use research like this?

People who project apocalyptic job losses are missing the point. So are people who say “you can’t trust that projection.” This transformation is a big deal, whether you’re in government, industry, or education. You need data and analysis to have an intelligent discussion about it. Smart people don’t swallow this stuff whole, but they do use it get smarter.

Next time you read a projection like this, don’t just goggle at the number. Ask yourself: is this big enough to affect me? What should I be doing about this trend? What should I be learning next? How can I dig into this?

Then you’ll escape the pessimist/optimist/realist trap. You’ll be thinking like an analyst

Photo: DSC_0126 via Flickr

Graphic: Forrester Research

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.