Career Prospects Post-AGI

Posted on Mar 2, 2025

Some catch-up reading and listening for various mavens conjecturing on the topic of work, especially in software, as we approach the AGI/AGEI “endgame”:

Not much here that’s unambiguously rosy. Cagan and O’Reilly provide probably the best optimistic cases. Yegge and Amodei less so in these particular articles. Amodei refers to another article, Machines of Loving Grace, where he presumably emphasizes the positive potential of the models and agents that he’s at the forefront of delivering.

Having listened to and read these thoughts over the past few days, there are a couple points that are sticking with me.

One from Tim O’Reilly’s article:

AI evangelist Ethan Mollick is also a fan of Bessen’s work. This is why he argues so compellingly to “always bring AI to the table,” to involve it in every aspect of your job, and to explore “the jagged edge” of what works and what doesn’t.

This dovetails nicely with Yegge’s insistence that CHOP is do or die for developers, and his introductory point that all workers need to learn to be LLM-first in as many of their job duties as possible.

The other is from Amodei’s interview with the Hard Fork guys:

I’m also doing more to take care of my health, but you should do that anyway. Right? I’m also making sure that I track how fast things are changing in society, but you should do that anyway. So it feels like all the advice is of the form doing more of the stuff you should do anyway.

In other words, this is really not the time to be screwing around. Or, more formally stated, the potential returns on fulfilling your responsibilities to yourself and to others are higher than they have ever been.

Yegge includes in his surveys an argument that humans will continue to operate on the “branch” nodes of any work tree. I would assume that LLMs will continue to push human involvement lower and lower toward the trunk, though.

It occurred to me while reflecting on all of this that the only thing that’s differentiated from LLMs in a robust way is being human and living the human experience. Of course LLMs can make assessments about the human experience based on infinitely scaled prediction context, but an LLM or agent can never have the equivalent intrinsic value of a living, breathing, experiencing human. It’s hard to imagine how the human variety of experience might be valued in economic markets, other than the well-worn path of creatives and their cults of followers.