I want to flag a concern of @Davidmanheim that the section on AI-directed labor seems to be relying too heavily on the assumption of a normal distribution of worker quality, when a priori a fat tailed distribution is a better fit to real life data, and this means that the AI directed labor force could be dramatically underestimated if the best workers are many, many times better than the average.
Quotes below:
(David Manheim’s 1st tweet) @rosehadshar/@Tom Davidson—assuming normality is the entire conclusion here, it’s assuming away the possibility of fat-tail distribution in productivity. (But to be fair, most productivity measurement is also measuring types of performance that disallow fat tails.)
(Benjamin Todd’s response) The studies I’ve seen normally show normally distributed output, even when they try to use objective measures of output.
Though I agree the true spread could be understated, due to non-measured effects e.g. on team morale.
(David Manheim’s 2nd tweet, in response to @Benjamin_Todd) When you measure the direct output parts of jobs—things like “dishes prepared” for cooks or “papers co-authored”—you aren’t measuring outcomes, so you get little variation. When there is a outcome component, like profit margin or citation counts, it’s fat-tailed.
(David Manheim’s 3rd Tweet) So for manual workers, it makes sense that you’d find number of items produced has a limited range, but if you care about consistency, quality, and (critically) ability to scale up, the picture changes greatly.
Interesting, thanks.
I wouldn’t have thought that consistency and quality were still capable of massive improvement for manual workers already in the top 10%.
Also, not sure it’s realistic to assume that people that today are unproductive would rise to the very top 0.1% even with ideal AI coaching.
But I do think the spread could be much bigger if AI is allowed to redesign factories from scratch.
There’s a lot of uncertainty here, even more than I’d realized
Would be interested in any evidence you have of fat tails for manual workers
And David’s reply:
I don’t have any evidence like that, but I also think that it wouldn’t show up in numbers that get collected. (I would argue that workers who have output that passes QA at linearly better rates must have exponentially better “quality” in a sense, but it’s not obviously true.)
I want to flag a concern of @Davidmanheim that the section on AI-directed labor seems to be relying too heavily on the assumption of a normal distribution of worker quality, when a priori a fat tailed distribution is a better fit to real life data, and this means that the AI directed labor force could be dramatically underestimated if the best workers are many, many times better than the average.
Quotes below:
Adding my reply tweet:
And David’s reply: