CEOs Who Think AI Replaces Their Employees Are Just Bad CEOs
from the that's-not-what-ai-is-for dept
In the last three months I’ve had people forward me four separate examples of a CEO losing his or her mind over AI. What’s been striking to me is the similarity in each case: It would be an “all hands” email in which the CEO talks up how amazing LLM tools are and saying that everyone in the company MUST start learning to use them immediately or they should look for a job elsewhere. Sometimes they talk about hiring “consultants” to come in and teach the team how to use the tools properly. Sometimes they are setting up “office hours” or internal “AI hackathons.”
But in every case the gist is the same “holy shit AI is amazing and you are expected to use it at your job all the time.” The worst case of these were the few companies that set up token leaderboards, which is perhaps the dumbest way possible to encourage learning how to use LLMs well. Good usage of AI includes learning how to view tokens as a scarce resource. Simply counting how much you use as a good thing is ridiculous because it’s incredibly easy to waste tokens on counterproductive uses.
As regular readers of Techdirt know, I actually do think that these tools are powerful and important, but I also think there are many problems with them and limitations to how useful they really are. I think when someone learns how to use them well and willingly chooses to use them as a tool to assist their work, they can be quite powerful. But the willingly choosing to use them part of that is important.
No one who is forced into using these tools will ever learn to use them well.
So CEOs losing their minds over the tech are not being helpful. Box CEO Aaron Levie — himself a genuine AI believer — puts his finger on exactly why.
CEOs are uniquely prone to AI psychosis because they’re sufficiently distant from the last mile of work that still has to happen to generate most value with AI.
So when they play with AI, they see the happy path results, often not considering the next 10 or 20 things that have to happen to get sustainable results from agents.
“Look I made this awesome product prototype”. Yes but you didn’t have to review the code before it went into production and fix a bunch of issues.
“Look I generated a contract”. Yes but you didn’t verify all the terms before it goes out to the counterparty and didn’t have to wire up all the past contracts to work with.
The best thing you can do as a CEO is to use AI a ton to figure out the real implications of agents in the enterprise, and come out the other side with an appreciation for both the upside and the real work that goes into them.
I will say that I hate the term “AI psychosis” because the term is extremely misleading, and many psychologists and psychiatrists have complained that it is inaccurate and may cause more problems itself. But the general sense that CEOs are going overboard with AI is definitely happening.
And I think Levie’s thinking as to why is also dead on.
Much of the issue may be in how disconnected the traditional CEO is from the people at a company actually getting stuff done. Normally, they have teams and layers and the actual work of getting things to work in a real way is so far removed from a CEO that they just get snippets of the details that filter back through the various org charts.
The problem tends to show up when a CEO is handed an agentic tool like Claude Code, and has it create something, which will work just fine, and thinks “oh, wait, why do we need so many people, when I can just sit here and make things work?”
This is a bad CEO.
Making things work is different than making things work well. Or well at scale. Or well at scale in a specific environment. Obviously, it depends on the kind of project and what it’s being designed to do, but oftentimes the reason a company has a bunch of employees is to fill in the seemingly small, but incredibly important details that CEOs might not ever get much visibility into: things like security or legal compliance or accessibility or who knows what else.
Using an agentic tool to build something that works is all well and good, but building a product for the mass market to use — and use well, and use safely — involves much, much more. Agentic coding tools can sometimes help with that too, but the leap from “I built a thing” to “therefore anyone can build a thing” misses the entire point of why you hire knowledgeable, experienced people in the first place. It’s also why I think the best case of these tools is building totally personalized tools to assist you in accomplishing a specific task, and not for building mass market tools.
This all reminds me of cargo cult thinking: The CEO knows that somewhere in the org, employees are pecking away at computers and work gets done. So they figure that themselves pecking away with Claude Code and seeing work get done is the same thing. It’s not. All those other steps those people are handling — the ones the CEO never sees — still need to happen.
That’s not to say employees wouldn’t benefit from a deeper understanding of both the power and the limits of these tools — they would. But there’s something darkly comical about watching a CEO go all in on the tech and then immediately conclude it means they can fire half the staff.
It seems pretty clear to me that companies that think they’ll be able to layoff huge swaths of workers because of LLM tools are going to find out they’re mistaken pretty quickly. The power of LLMs is that when used well and used willingly it can help employees to get more done, but that doesn’t mean you need fewer humans. You need more humans who know how to work productively.
Separately, companies pointing to LLMs as a reason for large layoffs are, in most cases, just using it as an excuse. They over-hired, and “AI efficiencies” is a much more palatable story for Wall Street than “we made bad headcount decisions.”
Levie’s prescription, though, is right: CEOs should learn how the tech works, but that includes the limitations of the technology. If a CEO thinks the prototype they vibe coded is production-ready, let them ship it and see what happens. If they think a vibe coded contract is as solid as one a lawyer reviewed, let them find out what the legal bills look like when it falls apart.
Yes, the tools are powerful, but a CEO who thinks they replace the work of employees is simply a bad CEO.
Filed Under: aaron levie, ai, ceos, llms, work
Comments on “CEOs Who Think AI Replaces Their Employees Are Just Bad CEOs”
“Yes, the tools are powerful, but a CEO who thinks they replace the work of employees is simply a bad CEO.”
I’d take it a step further: Such a CEO feels self-important and doesn’t really like their employees in the first place.
I’ve worked with upper management at small-to-medium sized companies, so I don’t know if the same applies to big, public, fortune 500 sized firms, but in my experience, nobody is thinking about how the business will work in 2 years, let alone 10.
Everybody loves making long-term projections; it’s easy to make them look good. But nobody is looking back more than 13-15 months, so they know that whatever they put in these rosy projections will be forgotten in a few board meetings.
If they’re told that AI can provide 50% of the productivity for 10% of the cost of an employee, they do layoffs, and let the business coast while raking in the profits for a few years. When business starts to suffer, nobody does an autopsy of how they got there. Or it’s a Private Equity owned business, they really don’t care, since it will have been sold off to the next sucker in the meantime.
All that is to say, many CEOs are incentivized to lie and make employees suffer much more than they are incentivized to build a sustainable business model for more than 5 quarters (comparing “same quarter, last year”) at a time.
So the problem (and this existed before LLMs), is that there’s a paper-thin margin between your “Bad CEO” and a “Normal, self-interested CEO just doing exactly what the Board wants and expects.”
My anticipatory frustration is that it will probably take significant failures with catastrophic financial and possibly economy-scale destruction before the lesson is learned higher up, whether it’s CEOs learning or boards/investors jettisoning CEOs who haven’t learned, and then it’ll still take time for lower managers at smaller organizations to get the idea that maybe “AI” isn’t a solution to everything. The only thing that seems to mitigate the possibility of a massive awakening is the fact that our society is inured to the idea that incidental tragedy and even intentional cruelty is something we can happily tolerate as long as our lives aren’t overly disrupted. We’ve accepted mass shootings, insurance denial fraud, wartime gas prices, ICE/CBP stealing our neighbors in broad daylight. What more is an LLM bot messing up our prescriptions or signing off on machine parts that are going to cause accidents?
I’d say we need to take Congress and the White House away from the profiteering regressives, but I doubt Dems will much better at regulating the issue with any amount of real insight into its nature. California trying to big brother everyone’s 3D printer is a good indicator that they’ll just take a bumbling “nerd harder” approach.
Ultimately the the best positions to replace in a company with AI
The best positions to replace in a company with AI is the CEO and other C* positions.
They do NO physical labor, they cost thousands of times more than other workers, they basically follow the boards instructions (or get removed), An AI with the proper training, a strict logic tree and and as much ethical subroutine as the board would allow would make better decisions than the C* of most corps have over the last 50 years.
Re:
As someone who has met hundreds of C-level executives at companies of all sizes, including a number in the Fortune 500, I heartily second this. I have yet to meet even one such person who impressed me as intelligent, educated, literate, and most importantly, ethical. American business has structured itself to reward the grifters, the bullshitters, the climbers, the backstabbers, etc., and has done this so effectively that it’s almost impossible for a qualified and decent person to get to the top.
So, YES, let’s replace every C-level executive at these companies with AI. It can’t possibly be worse — which given how shitty contemporary AI products/services are is saying a lot. But it’s true.
Re: Re:
All CxO level positions make more sense if you see them as salespeople and marketers. They don’t create value, but are still necessary in some capacity to turn the value into dollars.
Whether they are selling end-user product, stock, or even more nebulous things like “consumer confidence,” they do have a skillset that isn’t simple to replace with AI. That’s not to say they aren’t, on aggregate, vastly overpaid. Just that the caricature of the old guy on perpetual vacation while collecting millions per year is not real.
Re: Re: Re: they do have a skillset that isn’t simple to replace with AI
No, they don’t. They are salespeople.
Perfect for replacement with an LLM.
Re: Re: Re:
Selling end-user product is the marketing team’s job. Selling stock is only a result of the fucked up market system we have where the stock price doesn’t exist as a means of supporting a valuable company but as a profit-driving vehicle, so any skill here is like saying you have a skill in convincing workers to accept less than their labor is worth. It’s a “skill,” but it’s not a good skill or one any moral person should exercise. In a rational system with safeguards against gaming the market, stock should sell itself.
A coinflip could probably do better as far as CEO decision-making skill is concerned since just raw mathematical chance has no fragile ego or short term destructive profit motive or exploitative authoritarianism or illusions of grandeur or childhood trauma. Many CEOs have run functional companies into the ground and cashed out in a system of perverse incentives. That’s worse than a hallucinatory LLM would do unless the LLM was trained only on yacht club and Epstein Island conversations and Musk tweets. Imagine investing all that CEO pay, stock options, and golden parachute funds in the employees of the company instead…
Re: Re: Re:2
Shit, that’s pretty much what every venture capital firm was built to do.
Re:
An large language model is not actually intelligent, hence not “A.I.”. But perhaps the people buying into this puffery are not so intelligent either, and could be replaced by LLMs.
Similarly, 30 years ago, this type of CEO was probably bringing in management consultants to spout buzzwords and fire some people—never the CEO, though, and wouldn’t it be hilarious if a consultant noted that the CEO hired them without any real preparation or even a plan for what to do with the information? (“In conclusion, this report was a waste of company money…”)
This comment has been flagged by the community. Click here to show it.
Considering how amazing and great techdirt thinks AI is? Yeah, no surprise that’s how ceos react.
People who know absolutely nothing will think AI is amazing, and no one is dumber than a ceo.
Re:
That’s not the case. In general Techdirt has a very nuanced view of it. You would know that if you’d actually read the very article you’re commenting on.
Finally.
Re: Finally
A cock has come back to techdirt.
My company went from the C suite saying devs need to use AI from the start of their dev work and not just when they get stuck to we need to closely monitor how many tokens people are using. The reason? On June 1 GitHub CoPilot went from a subscription model to a per token model and the cost per token has basically gone up 27 times from the May token cost. And we also learned we are still in a “trial” period for some months and after that the price will go up another 3 times.
The devs have a cost limit per month they could use and rarely was it reached before, so far in June 100’s of devs have already reached their monthly quota and leadership is panicking.
Suddenly their is a lot more examining going on for AI use cases now rather than the “you need a reason you aren’t using AI” that existing previously.
We have one c-suite person like that, “use it 10 times a day or you’re not using it enough!”
I’m not sure his example-setting of shitty comics and 80s street-fighter-style animations are really leading the way though.
Malls
I remember during the tech bubble there were stores at the mall that would do web searches for you. Of course, they all went catastrophically out of business when the bubble burst. I see the obsession with AI experiencing a similar arch. Nobody will be making a profit using AI any more than they make a profit using web search. People will learn how to use AI as a tool to do their job and that is what will continue to make money.
Same ilk
the leap from “I built a thing” to “therefore anyone can build a thing” misses the entire point of why you hire knowledgeable, experienced people in the first place.
You know the managers who have to have their underlings in sight at all times and can’t manage employees who work from home because they don’t know how to measure productivity and output? Those CEOs are cut from the same cloth.
No. The CEO should ask some experienced engineers to evaluate the new tech, collect research findings, estimate ROI, and then decide if they should fund it.
You don’t expect the CEO of a hospital to do a bit of heart surgery to learn about a new piece of operating tech. And you don’t expect the CEO of a software company to do some vibe coding to learn how AI works.
Re:
“Learn how tech works” does not mean becoming an expert at utilizing that “tech”. But a hospital CEO should have a decent understanding of the top surgeries performed there, which will likely include some cardiology.
And, in fact, a lot of CEOs did work their way up from positions such as heart surgeon or programmer. Some people feel that promoting from within is the best way to find a good manager.
If we just want someone who can farm the work out to others, collate the data, and make a decision… doesn’t that seem like something a computer can do? If we’re gonna pay a human five or a thousand normal salaries to do it, we should have higher expectations.
While I enjoy watching CEOs pratfall, I do think there’s some things that are still worrisome.
You can need more humans who know how to work productively, while still ending up with less humans overall. If you have a set amount of work, if you make people more productive, you will need less people overall, generally speaking. There can be exceptions, like if something (like review) is a hard bottleneck, but not everything will be.
The issue CEOs are running into is overcorrecting and thinking that means you don’t need any people. But if it took you 100 employees with 25 of them reviewing to get something done before, it might only take you say, 75 now with 35 of them reviewing to get it up to snuff. But you’re still down 25 jobs, net. There are going to be a lot of jobs that won’t need 100 reviewers- that’s the point of increased productivity. (numbers are just for example).
The main question is if there’s induced demand- either the same 100 employees can get more done out of the same inputs, or it allows them to expand into new things they couldn’t do before. We don’t yet know how much of that there will be.
The biggest thing is that this should be a warning- CEOs want to replace your jobs, and will if they can. Labor needs to start locking in concessions now, while they have bargaining power. We also need to keep an eye on the rate of improvement- while they still need review, even compared to a year ago, they need much less of it.
Somewhat separately from all of the above, not all skillsets are transferable- the skillset that made someone a good code monkey in the past doesn’t necessarily translate to being a good reviewer, which tends to be more like a manager role. That will lead to some pain all on it’s own.
Re:
Long-term, I don’t really see how that can work. Either we need people to do jobs, or we don’t. If this replacing-workers-with-computers thing isn’t total bullshit, someone will be doing it in another country, or even just a company not bound by the “bargains”.
It could take a decade or a century for the “dinosaurs” to die out, perhaps delaying the problem long enough to solve it properly. Don’t assume that negotiating a good contract will actually resolve the problem.
At my org a VP was on stage telling the department that he just “vibe coded” a “Barbie dressing app” with his daughter over the weekend and it was “so easy that every engineer in this audience ought to be doubling their output immediately.”
What he doesn’t realize is the experience he had with the LLM: few technical constraints, zero integrations, no compliance audits, no platform interoperability concerns, no security frameworks….
The hard part of writing code isn’t the generating of code. It’s everything else: requirements, communication, making and conveying decisions, political persuasion, institutional knowledge transmission. Maybe most of all: good judgement.
Re: A new slang proposal...
Anytime a large company’s software works very poorly, it should be referred to as ‘vibe coded.’ e.g. “Adobe’s online license system was clearly vibe coded.”
To paraphrase an old adage: CEOs can remain bad longer than you can remain solvent.
Refining governance
Levie’s point is spot on, but I would refine it slightly. The further a CEO sits from the people actually doing the work, the more opaque the details of that work become. AI just makes this worse. Its output, on initial inspection, appears “finished”, which gives a CEO, who is already well insulated from the real work being done, one more reason not to look for it.
In my experience, I spend considerably more time reviewing AI output than I would reviewing the same work from a human. With a human I need to focus on the logic, coherence, and accuracy of their work. With AI, I get a polished result, but I have to dig deep into the topic to evaluate the veracity of its work – verify assertions, confirm it has done what i want rather than what I asked, validate references etc. To reuse a phrase from a piece I recently wrote – I need to handle the consequences of “Pattern recognition dressed in the language of expertise”.
Anyone who treats AI output as a finished product rather than a first draft is skipping this essential oversight step entirely. Executives who actively strip away essential controls are neglecting their role in corporate governance.
I’ve been writing about what happens to organizational judgment when oversight over technology degrades gradually rather than all at once at theyellowfox.net if it’s of interest.
We might be under a faulty base assumption...
Are we really sure that there is such a thing as a GOOD CEO in the first place?
It's a mystery...
Techbros: AI is the new digital messiah that can solve everything for cheaper than paying for those filthy ’employees’ to do it and do it quicker as well, so we need to use AI as much as possible on as many things has possible.
People get fired, stuff breaks and/or is more cumbersome, and the quality of products/services tanks because AI is putting out broken stuff that needs humans to fix it after the fact
Techbros: Why do all the filthy peasants hate AI so much?!
Another bit of the pendulum swing.
German court finds Google liable for errors in AI Overviews