Techdirt: CEOs who think AI replaces employees are just bad CEOs. The thread asks who's more replaceable
The argument: executives see a working prototype, the 'happy path,' and never the security, QA, and compliance work that ships it. Box's CEO calls it 'AI psychosis.' The thread had a sharper target.
The case is that bosses cut staff because they can't see the unglamorous work that turns a demo into a product, then blame AI for layoffs that were really over-hiring.
Several commenters flipped it: if a job is just relaying information, the most replaceable role in the building is the CEO's.
A dissent worth keeping: by the only metric that judges CEOs — keeping shareholders happy — AI-driven cuts can be 'good' CEOship, which is the whole problem.
The argument: cutting staff is a visibility failure
Techdirt's piece argues that an executive who believes AI can replace employees is advertising bad leadership, not reading a technological inevitability. The mechanism is a gap in what executives can see. They watch a working prototype — the "happy path" — and conclude they can cut headcount, because they have no visibility into the security, legal compliance, accessibility, QA and code review that actually make a product shippable. The piece quotes Box CEO Aaron Levie on executive "AI psychosis," the habit of ignoring "the next 10 or 20 things that have to happen to get sustainable results."
It also names specific practices: "token leaderboards" that reward wasteful AI usage, and all-hands emails demanding immediate AI adoption under threat of job loss — which, the author argues, only produces unskilled, resentful users. Layoffs blamed on AI, in this reading, are usually cover for earlier hiring mistakes.
The inversion: the most replaceable seat is the corner office
The thread's favorite move was to turn the headline back on its subject.
"Any CEO who wants to replace jobs with AI should first have to replace their own assistant with AI … I think we'd get zero volunteers from CEOs who have assistants." — a commenter relaying a challenge to executives
The metric-leaderboard critique got a first-hand second:
"lots of places ressurected Lines of Code as a productivity metric … Companies instituted token usage leaderboards. We can't stop that thing. It has too much momentum." — the same engineer from the SEV-0 cleanup thread
The dissent that makes the piece uncomfortable
The strongest pushback didn't defend the CEOs — it questioned whether "good CEO" even means what Techdirt wants it to mean.
"There is one and only one measure of whether a CEO is good or bad: Does the CEO keep the majority of shareholders happy? … When your domain of knowledge considers Jack Welch to be a genius, there is no floor." — a commenter on the shareholder-value standard
If that's the scorecard, AI-driven cuts that lift the stock are good CEOship by definition — which is precisely the trap the article is circling.
The slow-displacement history lesson
One commenter reached for a useful analogy against easy optimism in either direction:
"The working horse population continued to grow for almost 20 years after the introduction of the Oldsmobile in 1901 … didn't fall below the 1900 level until 1930." — a commenter offering the horse-displacement parallel
Displacement was real, and also took three decades. The thread's lesson isn't that AI changes nothing or everything. It's that "the demo works" and "we can fire people" are separated by all the labor an executive never sees — and a scorecard that only counts the share price.
Get breakdowns like this every day
The top Hacker News stories, unpacked — by email, Telegram, or RSS.
Apollo's chief economist says there's no AI jobs crisis. The thread says look at who's actually hiring
Torsten Slok points to job openings back above one per unemployed worker and 172,000 new May payrolls as proof ChatGPT isn't replacing anyone. Commenters say nearly all of that growth is one sector.