How many of these jobs will have a 15% or more drop in employment plausibly attributable to AI by 2031?

post by tailcalled · 2023-02-12T15:40:02.999Z · LW · GW · 5 comments

This is a link post for https://manifold.markets/tailcalled/how-many-of-these-jobs-will-have-a?r=dGFpbGNhbGxlZA

Contents

5 comments

I made a Manifold market on the plausible impact of AI by 2031:

5 comments

Comments sorted by top scores.

comment by Dagon · 2023-02-13T19:41:37.734Z · LW(p) · GW(p)

How does it map to a probability?  It states "...the market resolves to the count of how many of the following types...", but I can't see anything count-based.  What does a bet of "yes" or "no" actually mean?  Also, should you weight these jobs in some way - Software Developer, for instance is currently 1.425M, projected to go to 1.796M, where Chief Executive is less than 1/4 that size, and expected to decline.  Also, do you need some process for handling changes in the overall list (say, the list of "software and web engineers and testers" changes, so "software developer" has a different meaning or is removed entirely)?

I'd be kind of interested in a market for the top line - will BLS reported employment for 2031 be above or below the current projection of 166.452 Million.

Replies from: tailcalled
comment by tailcalled · 2023-02-13T19:56:19.401Z · LW(p) · GW(p)

What does a bet of "yes" or "no" actually mean?

Yes = the fraction will be higher than the current %, no = the fraction will be lower than the current %.

Also, should you weight these jobs in some way - Software Developer, for instance is currently 1.425M, projected to go to 1.796M, where Chief Executive is less than 1/4 that size, and expected to decline.

It is possible to weight things, but it is also possible to not weight things. I chose to not weight things.

Also, do you need some process for handling changes in the overall list (say, the list of "software and web engineers and testers" changes, so "software developer" has a different meaning or is removed entirely)?

Hmm added this:

An exception is made if the list of job types changes, e.g. jobs are split, renamed, or deleted. If the job types change due to AI fundamentally changing the nature of the job, then I will likely count that job as having a drop in employment, unless doing so would be silly (e.g. the nature of the job hasn't changed very much, more people are working in the job than before). If the job types change for other reasons, I will try to find a comparable job type in the new list to make comparable resolutions. If all else fails, I will likely remove the job from the list as above.

comment by Jeff Rose · 2023-02-12T15:56:24.219Z · LW(p) · GW(p)

"An exception is made for jobs that fail to reach their employment due to some clearly identifiable non-software-related shock or change in trends, such as an economic crisis or a war. Such jobs will be removed from the list before computing the fraction."

But macroeconomic or geopolitical events such as major recession or war are likely to affect all job categories.   So the correct way to deal with this is not to remove such jobs but to adjust the fraction by the change in overall employment. 


 

Replies from: tailcalled, tailcalled
comment by tailcalled · 2023-02-12T17:13:08.697Z · LW(p) · GW(p)

Changed it to:

An exception is made if there is a non-AI related factor that causes a widespread long-term change in the economic trajectories, e.g. a war. In that case I will attempt to adjust for this factor by changing the resolution criterion, e.g. by counter from the projected employment after the event happened.

An exception is made for jobs that fail to reach their employment due to some clearly identifiable non-software-related shock or change in trends, such as a job-specific crisis or change (e.g. if motor vehicles get banned for safety or pollution reasons). Such jobs will be removed from the list before computing the fraction.

comment by tailcalled · 2023-02-12T15:58:45.262Z · LW(p) · GW(p)

True, I will look into changing this criterion.