I previously posted on the jobs that I think are most vulnerable to artificial intelligence (computers, robots). Here's a table with the number of workers (in millions) in each of the top 15 job categories in 2012, and my predictions for what the employment will be in each of those categories in 2024, 2034, and 2044. I think the two jobs most vulnerable to artificial intelligence are cashiers and truck drivers; I expect about a 90% reduction in their numbers from 2012 to 2044. Other jobs I see as particularly vulnerable are bookkeepers/accountants and material movers (e.g. loading dock workers). But I don't see even one of the top 15 job categories in 2012 as gaining workers by 2044. And overall, I expect about a 50%(!) drop in the number of people in the U.S. employed in the top jobs of 2012 by the year 2044.
Robin Hanson recently had a blog post with the title, "This time isn't different"...relating how previous predictions of job losses due to computers and other machines didn't come to pass. I beg to differ. I think this time most certainly is different. That's because this time we're talking about computers/robots very shortly (in the next 1-2 decades) generally equaling then surpassing the capabilities of a human brain and body.
| Popularity, 2012 | Job | Employment relative to 2012 (i.e., 2012 = 1.0) | Percent Change, 2012 - 2044 | ||||
| 2012 | 2024 | 2034 | 2044 | ||||
| 1 | Retail Salespeople | 1 | 0.95 | 0.48 | 0.29 | -71 | |
| 2 | Cashiers | 1 | 0.88 | 0.53 | 0.12 | -88 | |
| 3 | Office Clerks | 1 | 0.93 | 0.71 | 0.50 | -50 | |
| 4 | Fast food prep and service | 1 | 1.04 | 0.89 | 0.59 | -41 | |
| 5 | RNs | 1 | 1.00 | 1.00 | 0.88 | -12 | |
| 6 | Waiters/waitresses | 1 | 1.04 | 0.78 | 0.52 | -48 | |
| 7 | Customer Reps | 1 | 1.05 | 1.00 | 0.91 | -9 | |
| 8 | Material movers (e.g., loading docks) | 1 | 1.00 | 0.62 | 0.14 | -86 | |
| 9 | Janitors | 1 | 1.05 | 0.57 | 0.19 | -81 | |
| 10 | Stock clerk, order filler | 1 | 1.16 | 1.26 | 0.84 | -16 | |
| 11 | Secretaries | 1 | 1.06 | 0.94 | 0.83 | -17 | |
| 12 | Bookkeeping, accounting | 1 | 1.00 | 0.56 | 0.22 | -78 | |
| 13 | General managers | 1 | 1.00 | 0.94 | 0.88 | -12 | |
| 14 | Tractor trailer truck drivers | 1 | 1.06 | 0.56 | 0.06 | -94 | |
| 15 | Elementary school teachers | 1 | 1.00 | 0.87 | 0.53 | -47 | |
| Totals | Top 15 jobs in 2012 | 1 | 1.00 | 0.76 | 0.48 | -52 | |
Postscript (11/10/15): I should have made it clear that all of these predictions are population-weighted. For example, I have employment in 2044, relative to 2012, for all top 15 jobs, as 0.48 (i.e., a 52% reduction in the number of total jobs in all 15 sectors). But that's a population-weighted number. For example, suppose the total U.S. population increases by 20 percent from 2012 to 2044. Therefore, a population-weighted value of 0.48 would actually look like 1.2 x 0.48 = 0.58. So if the population increases by 20 percent, the absolutely number of jobs would be 0.58 times the 2012 value, rather than 0.48 times the 2012 value. Of course, if the population didn't change at all from 2012 to 2044, the 0.48 value would be appropriate to calculate the total number of jobs relative to 2012.
Mark,
Perhaps you should update your exponential economic growth chart, since your prior analysis is a decade old...
I think the linear regression of 1800-2015 is the correct projection, since AI is what makes that extrapolation happen in the first place. I don't think AI makes it go faster than that (the Arnold Kling projections, and yours), but rather AI enables the still-fast linear regression.
At any rate, vast prosperity is just 15-25 years away from now (now being 2015).
Posted by: TFH | May 11, 2015 at 10:26 PM