The Core

I think for the first time ever, I'm going to say that that was not a "must read" for me (though I read the whole thing).

The main surprise in it for me was learning that Shopify is 35% the size of Amazon in gross revenue that it supports through its platform, but it wasn't clear to me what the connection to AI was in most of the slides in that section.

Check out the Understanding AI blog. I think it generally goes deeper and farther.  I would say almost every post I read teaches me something new. I've just discovered it, so that may wear off. And, yes, the posts are really long, so they are not quick reads.
https://www.understandingai.org/

>> how long

How about this for a guess?

Level 1: If your job is some form of moving electrons, then the AI is coming for you.

All these people who insist that "An AI can't do my job," are hosed. An anesthesiologist friend reacted that way when I asked ChatGPT to suggest anesthesia for a surgery and shared the results.

Most people don't understand that what they call skill is either 1) advanced pattern matching or 2) precise movement. Both of those things will unquestionably be performed better by machines at some point. When that point comes is partly a function of how hard the task is, how different it is from an existing task, how disastrous mistakes are, and how much money can be saved by removing the human from the loop.

So for jobs where the pattern-matching looks a lot like things that AI is doing and the consequences of mistakes are low (e.g. writing marketing copy), most of those jobs disappear in the next few years, being done by the infinite interns.

Law and medicine and engineering will have to have humans in the loop for many years to come if for no other reason than to have someone to sue. If I were 50 and in one of those fields, I would expect to survive to retirement with a job. If I were 40, I would not.

Already, it's clear that Waymo cars are safer and have fewer accidents than human-driven cars.

Level 2: If your job is some form of moving molecules, then it depends. That's much like level 1 in terms of the key drivers.

3: If your job depends on *directly*(1) "moving" people physically or emotionally, then things look better (i.e. coach, infantry officer, massage therapist, counselor).

(1) meaning, not counting bus drivers or elevator operators, but humans who "touch" humans physically or emotionally. But I think having people with MDs who sit in an office, take history and order tests is done. It should be done within 10 years, but due to big lobbies and regulatory capture, I expect much longer. Though, honestly, if Elon and Vivek had 10 years to work at DOGE, I would expect this to be something that they would attack given their background and biases.

https://www.understandingai.org/p/seven-big-advantages-human-workers

You like verbose, I like info bullets, EG.

I think the usage will remain roughly along the industry categories in this chart (except the lawyers, they're lying and already using it a lot).  But the penetration rates will skyrocket.


 

>lawyers

Trail lawyers are are a different category.  Think in British terms of barrister and solicitor.  A good trial lawyer (barrister) argues in front of a jury and has a personality that can connect, and communicate with the humans on the jury.  This is different from a solicitor who fills out the proper forms. (I'm over simplifying.)

For example, a successful trial lawyer once told me that his undergraduate majors were English and theater and those skills were invaluable as a trial attorney.  You have to be able to command the language and act in front of a jury.

That said:  All lawyers have a reality distortion field, all of them. Trial lawyers (barristers) are able to turn it up to 11.

This is maybe the third post I've seen along these lines in the field of exercise physiology: