Brynjolfsson is a very good economist, and yet I have found he is inclined to exaggerate the effects of AI. Here is an article from the BLS from 2022, which looks IMHO excellent, demonstrating that some jobs that had been selected as likely targets for shrinking by AI, nevertheless grew as a result of factors the economists did not take into account. I remember all the hooplah about MOOCs a number of years ago, which proclaimed that teaching was a vanishing profession -- which turned out not to be true. Mr Trump's disruption of the BLS will have incalculably large effects on the production of great economics papers by people who do not splash their famous names across papers, but somehow produce excellent work. https://www.bls.gov/opub/mlr/2022/article/growth-trends-for-selected-occupations-considered-at-risk-from-automation.htm
I'll pressure you the same way that I did the Dr. As. The unemployment rate for recent college grads went above the unemployment rate for all workers *before* the pandemic, which was before widespread knowledge of LLMs and AI agents.
I'm not dismissing that AI has had an outsized influence in the past few years, but what trends might we be overlooking from before 2022?
So, while the pre-trend is of course concerning, my fear is that AI (broadly defined) may make it harder for the Fed to effectively address part of its dual mandate.
I think my previous point was not clear. To be clear, my position is that AI is accelerating the issues the labor market has with 1- Gen z 2- value of higher education. Add that to increasing uncertainty, new tech. Industry has a scapegoat in AI. While pre pandemic might have shown signs, the trend has been more aggressive and sustained.
Higher education needs to do better with communicating its value proposition of better yet improve its value. Thanks for the “pressure” it helps hone the message. I will leave the rest to Brandon
The gap between all workers and recent college grads has been falling pretty consistently since 2012. Based on just the gap, it actually doesn't seem like AI is accelerating this decades-long trend, but it is a pretty nice scapegoat.
I think we're seeing the knock-on effects of how the 2008 Financial Crisis impacted young adults at that time. Many of them are now the parents of these "recent college grads."
There was (and still is) a big push for students to get "marketable degrees" so that they can avoid what happened to their parents in 2008. Students went from learning skills to instead completing checklists to get a degree. When I ask incoming students why they picked the major, 95% of the time the answer is "to get a good job."
It feels like this is part of a long-term pivot away from human capital acquisition toward signaling theory, and the market has gotten better about screening out people who haven't acquired skills or who don't know how to acquire skills.
I agree with this hypothesis. If the only reason you are getting a degree is for the sheepskin effect, then the market is identifying that the candidates don’t have the skills necessary to succeed.
I also think it helps with downplaying the role of uncertainty in today's market. Obviously, that doesn't explain the decade-long decline in the gap, but it does make it easier not to blame current policy for making things more complicated.
Nice work Brandon. You point out some very truthful and unfortunate situations. Young people that has already invested a lot of time into just getting to the starting line cannot start the race. This will have a greater impact than we can imagine. My biggest worry is that the system will be too slow to react and it will solidify the losses.
Brynjolfsson is a very good economist, and yet I have found he is inclined to exaggerate the effects of AI. Here is an article from the BLS from 2022, which looks IMHO excellent, demonstrating that some jobs that had been selected as likely targets for shrinking by AI, nevertheless grew as a result of factors the economists did not take into account. I remember all the hooplah about MOOCs a number of years ago, which proclaimed that teaching was a vanishing profession -- which turned out not to be true. Mr Trump's disruption of the BLS will have incalculably large effects on the production of great economics papers by people who do not splash their famous names across papers, but somehow produce excellent work. https://www.bls.gov/opub/mlr/2022/article/growth-trends-for-selected-occupations-considered-at-risk-from-automation.htm
I'll pressure you the same way that I did the Dr. As. The unemployment rate for recent college grads went above the unemployment rate for all workers *before* the pandemic, which was before widespread knowledge of LLMs and AI agents.
I'm not dismissing that AI has had an outsized influence in the past few years, but what trends might we be overlooking from before 2022?
Great catch! I think the short answer is "we don't know." I think Dr. A is right that AI is accelerating whatever pre-trend existed, and I think those trends probably have a few underlying causes. Check out this very smart post describing some potential reasons: https://open.substack.com/pub/findingequilibriumfuturehighered/p/an-employment-crisis-for-new-college?r=1o70di&utm_campaign=post&utm_medium=web
So, while the pre-trend is of course concerning, my fear is that AI (broadly defined) may make it harder for the Fed to effectively address part of its dual mandate.
I think my previous point was not clear. To be clear, my position is that AI is accelerating the issues the labor market has with 1- Gen z 2- value of higher education. Add that to increasing uncertainty, new tech. Industry has a scapegoat in AI. While pre pandemic might have shown signs, the trend has been more aggressive and sustained.
Higher education needs to do better with communicating its value proposition of better yet improve its value. Thanks for the “pressure” it helps hone the message. I will leave the rest to Brandon
I went ahead and pulled the data from the NY Fed to get a better sense of when this trend started: https://datawrapper.dwcdn.net/mzVA9/1/.
The gap between all workers and recent college grads has been falling pretty consistently since 2012. Based on just the gap, it actually doesn't seem like AI is accelerating this decades-long trend, but it is a pretty nice scapegoat.
What’s your hypothesis for the cause?
I think we're seeing the knock-on effects of how the 2008 Financial Crisis impacted young adults at that time. Many of them are now the parents of these "recent college grads."
There was (and still is) a big push for students to get "marketable degrees" so that they can avoid what happened to their parents in 2008. Students went from learning skills to instead completing checklists to get a degree. When I ask incoming students why they picked the major, 95% of the time the answer is "to get a good job."
It feels like this is part of a long-term pivot away from human capital acquisition toward signaling theory, and the market has gotten better about screening out people who haven't acquired skills or who don't know how to acquire skills.
I agree with this hypothesis. If the only reason you are getting a degree is for the sheepskin effect, then the market is identifying that the candidates don’t have the skills necessary to succeed.
I also think it helps with downplaying the role of uncertainty in today's market. Obviously, that doesn't explain the decade-long decline in the gap, but it does make it easier not to blame current policy for making things more complicated.
Nice work Brandon. You point out some very truthful and unfortunate situations. Young people that has already invested a lot of time into just getting to the starting line cannot start the race. This will have a greater impact than we can imagine. My biggest worry is that the system will be too slow to react and it will solidify the losses.