How you’re turned into an appendage of a hallucinating machine, with no right to object
A year and a half ago, I quit one of the largest international consulting firms. During my final year there, I already sensed that something fundamentally wrong was beginning to happen. There were constant recommendations to use ChatGPT — even in situations where it wasn’t needed and without any clear policy on how or why. This quickly began to turn into pressure.
And now I realize that was just the mild part. I can very well imagine how those who are still inside the system feel right now.
It’s that crushing feeling, as if a massive steamroller is bearing down on you, and you can’t move from the spot where someone has chained you down.
That’s what a nightmare feels like. And there’s no greater relief than waking up from it.
The Layer Between Prompt and Hallucination
But those who remain there will not be able to wake up. They are forced to endure what is happening, knowing full well that it is nothing other than madness.
And this madness is now embedded in the corporate ethos. It has become the new normal, an article of the corporate constitution.
You are no longer hired to do work you can be proud of. You are hired to watch over the hallucinating machine and not dare to do anything on your own.
Your experience is now trash, your creativity is a systemic anomaly, your opinion… it simply isn’t part of the equation anymore.
You are no longer an engineer or an analyst. You are a buffer between the prompt and the hallucination.
A biological appendage to an algorithm that understands reality slightly better than a rock (though a rock doesn’t hallucinate).
Your usefulness is now measured by how many tokens you feed to the so-called “AI.”
This is called tokenmaxxing. Yet another product of AI psychosis, yet another repulsive corporate euphemism.
These are the very chains that have bound you to your workplace.
Slop as a Positive Efficiency Metric
Tokenmaxxing isn’t an absolutely new phenomenon. Back in the 1980s, management was already trying to measure the productivity of a process it didn’t understand.
All corporate minds could come up with was counting lines of code: the more you wrote, the more productive you were; the less — well, you get the idea.
The consequences were predictable: the code ballooned, solutions became more complex. Systems got worse and worse, while reports looked better and better.
Eventually, it became clear to everyone that, as Bill Gates put it, measuring programming progress by lines of code is like measuring aircraft building progress by weight.
In the end, common sense prevailed. But back then, it was much easier to admit the mistake and relatively simple to fix it. Now everything is much more complicated. AI hype has escalated into a psychosis, and questioning the effectiveness of AI has become a de facto taboo in the corporate world.
And of course, corporate management still considers its inability to understand production to be a virtue. So how are they really supposed to know that an effective engineer strives to reduce the amount of code, rather than inflate it?
And not just an engineer. It is obvious to any professional that brevity is the soul of wit. A professional simplifies, shortens, and removes the superfluous. As Michelangelo put it, “I saw the angel in the marble and carved until I set him free.”
The difference between a human and a machine is that a human, if they are a professional, knows in advance where to look for a solution. They know the direction even before writing the first line of code or making the first Google search.
They don’t need to read the entire internet. They filter out the unnecessary.
Can this creative process be measured by the number of tokens burned?
I don’t think so. But corporate management thinks differently. Its goal is to burn through at least the planned amount of resources. The more that’s burned, the more is required. The scale of the waste masks how pointless it all is, and the amount of slop produced signifies the level of AI adoption.
And this is everywhere now:
- Marc Benioff of Salesforce has implemented a “token counter.” Dashboards track AI usage in real time. If an employee hasn’t met their “interaction” quota with the algorithm by lunchtime, they end up on the “low-productivity” list.
- Microsoft has also implemented similar dashboards in the Azure and M365 admin panels. They show the “volume of AI resource consumption” per employee. Managers can now see how many tokens a particular department has burned. The more that’s burned, the better — the more successful the “AI implementation” is in the corporate workflow.
- Meta burns 60 trillion tokens a month. This costs them $100 million. And they need these numbers to grow, because investors must believe that AI is “taking over the world.”
Okay, here you might ask yourself: if this technology is really so “revolutionary,” why is this coercion even necessary? Why turn work into torture?
Why Are They Hiding the Truth?
The answer is simple: because in reality, the whole narrative about “hyper-productive AI” is falling apart. To call a spade a spade, it’s a lie. And of course, corporations need to hide this.
Look at the facts that corporate management diligently ignores, while its employees frantically burn tokens to avoid being thrown out onto the street for “low productivity”:
- The MIT report, The GenAI Divide (July 2025), showed that 95% of all corporate AI pilot projects failed. They yielded neither profit nor productivity gains. In most cases, they actually reduced productivity.
- Research by Carnegie Mellon University (September 2025) confirms: AI agents fail to handle even basic office tasks in 70% of cases.
- METR found (July 2025) that AI coding tools actually slow developers down (the study cites a figure of 19%). They spend more time fixing the machine’s “hallucinations” than they would if they were writing the code themselves.
- Finally, the Remote Labor Index report (October 2025) showed that AI agents can perform only about 4% of real commercial tasks typical of ordinary projects (game development, design, data analysis, animation).
Clearly, it’s not about statistically acceptable glitches, but about a recurring pattern, is it?
But for a board of directors, to acknowledge this scientifically proven reality would mean committing financial — and possibly career — suicide.
Big Tech stocks are overvalued by trillions of dollars thanks to faith in AI. If that faith disappears, the bubble will burst. If it bursts, then, as Sam Altman put it, someone is going to lose a phenomenal amount of money.
So tokenmaxxing isn’t just another idiotic approach. It’s a magic formula that turns obvious nonsense into a corporate-standard KPI. It’s the perfect smokescreen. When actual results approach zero, management shifts the focus to “adoption intensity.” To shareholders, a ‘token consumption’ graph skyrocketing looks like success. It seems to them that the company is rapidly transforming and that soon they’ll be able to get rid of the lion’s share of the staff.
In reality, the company is simply forcing this staff to simulate activity.
This is a fundamental deception. If the AI spits out nonsense, you spend tokens to fix it. If it makes a mistake again, you spend even more tokens.
From a common-sense perspective, this is a disaster. From a tokenmaxxing perspective — it’s a double success! Resource consumption is rising, the financials look great, shareholders are happy, and investors are opening their wallets.
So you must keep quiet and burn tokens. You don’t really want to jeopardize the market cap of the corporation that hired you, do you?
A Forced Romance with a Chatbot.
So, now work isn’t what it used to be. Now you’re forced to take a task that could be done in minutes and turn it into hours of humiliating back-and-forth with an AI, as if it’s the one solving the problem, not you.
You ask it to write code, edit its nonsense, ask again, edit again. You fake “synergy,” so your manager sees the usage charts go up.
As a result, senior engineers and lead analysts now spend up to 40% of their work time on so-called “prompt grooming.” This is nothing more than a way to make simple work as long, expensive, and noisy as possible. And if you try to object that “AI is just getting in the way here,” you’ll quickly be reminded: the use of AI is no longer up for debate. It’s a top-down directive.
And that means you’re obligated to be slow and stupid along with the machine. This is the essence of your transformation into a biological interface: you exist solely to legitimize the “integration of AI” into corporate operations.
You must accept that if they perform a virtual lobotomy on you, the corporation will lose nothing.
Conclusion
There is some good news, too. All this insanity doesn’t mean everyone has gone completely mad. Not at all. It’s just that everyone is playing their part in this bloody circus. And some of its participants are fully aware of it:
- The CEO, whose income directly depends on the stock price
- Employees who are experiencing “AI progress” firsthand
- Some investors who know when to enter and when to exit, or who have insider information
- NVIDIA and companies that prepare datasets for AI models
- Companies that maintain the infrastructure of the AI bubble, such as those building hyperscale data centers.
The bad news is that this show requires scapegoats. And these scapegoats are ordinary workers — the most vulnerable, powerless, and exploited category in this entire structure.
And if we ask a simple question — what does society get out of all this?
What do the people whose labor keeps everything running get?
The answer becomes painfully obvious.
The question is no longer whether we are ready to accept this as the new normal. The question is whether we are ready to admit that it has already happened.
