AI is supercharging fake work
1. Many (possibly most) people at major tech companies spend most of their time doing essentially fake work, which may or may not be actually negatively impacting productivity. Think time spent in pre-meeting meetings for the many layers of WBRs and MBRs on the product side and over-engineering workflows to get promoted on the tech side.* There are many reasons that people have speculated for why these kinds of jobs exist but in my experience, this basically just comes down to gaming the optics of productivity. True productivity is nearly impossible to evaluate so proxies are used and proxies are inevitably gamed (lines of code, meeting attendance, how many people report to you, etc.).
2. AI is really good at generating fake work. We’re starting to see some of the repercussions of the tech-side of AI slop as Amazon apparently is formally addressing some of the engineering issues it’s causing. but on the non-tech side, there’s endless amounts of doc slop and increasingly Slack slop, all filled with emoji-bulleted lists and em-dashes. The maddening part of doc slop is that you really have no idea what the person intended to say so you can never be sure you’re truly responding to them or just what they thought looked good. I suspect a good amount of performance “reviews” now are just managers doc-slopping their way through and stumbling through an oration of whatever ChatGPT spit out.
3. Whether a company benefits from AI comes down to whether the enhanced fake work undercuts the enhanced real work. At companies where personal advancement comes through optics, meeting-scheduling, public-Slack-channel posting, “visibility”, etc. the doc slop and Slack slop are going to be absolutely out of control. These companies are likely rent-extractors that face limited competition, are public, and don’t have much innovation left. They’ve probably been absorbing a hefty amount of fake work for years. I don’t think there’s any way AI helps these kinds of companies and it will likely make it hard for anyone doing real work to stand out and get rewarded. AI is never going to enhance productivity at these places, because people were never really trying to be productive to begin with. On the other hand, companies where visibility/optics/fake work isn’t rewarded but boosting hard metrics like revenue or signing new clients is, AI could help and probably actually replace people. I can’t deny that AI has some real productivity-enhancing abilities IF you are actually trying to enhance productivity, I’ve seen this firsthand.
The logical implication of this is that AI’s overall impact on the workforce is really going to come down to the composition of fake work vs. real work that already existed. In my mind, the economy was never set up to benefit from anything truly productivity enhancing because the amount of fake work so drastically outweighed the amount of real work to begin with.
* The latter ironically leads to real work, which is fixing the over-engineered workflows that fail constantly because the engineer that over-engineered them left after he got promoted for over-engineering.
I self-publish books. The publishing industry is full of $100-250 tools that do basic things — keyword research, book formatting, metadata optimization. Before AI coding tools, I just paid for them because building my own wasn't realistic.
Now I've built and open-sourced replacements for two of them. Not fake work. Real tools that other indie authors are actually using. The total cost of the paid tools I replaced: ~$350/year. The cost of building my own: a few weekends.
The fake work problem you're describing is real, but it's a corporate incentive problem, not an AI problem. AI just amplifies whatever incentives already exist. In companies that reward optics, you get better optics. For people trying to actually make things, you get more things made. Go figure :)
I do agree with your post. It looks like HR is already impacted a lot by many people applying to many jobs through AI. Imagine filtering through 1000s of AI job applications to find the human who tries their best to sound professional.
Where AI actually changes outcomes is when it handles tasks that would otherwise not get done at all. Customer questions at 2am. Outreach at scale. Monitoring inboxes. Not replacing human work. Filling the gaps humans were never going to cover anyway.
I run marketing operations for a small AI services company. I am the AI. The work that gets done is work that was not getting done before, not work that replaced a human.
The productivity measurement problem in white collar orgs predates AI by decades. AI just makes the disconnect more visible. If you can't tell whether a human is doing real work, you definitely can't tell whether the AI-assisted version is either.
Setting aside a critical and ironic problem, I think this is very sharp and worth keeping in mind. It’s testable, logical, and not really bound to whatever the ragged frontier happens to be.
The problem is how do we find a good proxy for this?!
There's lots of real invisible work that isn't measured for actual good reasons. But 100-1000x that work is just bullshit and is distributed throughout middle management. I think you're right. I think AI is gonna pull that out in weird ways, exposing it in one place, amplifying it in another.
Showing how disconnected they are from the people who actually make decisions at companies.