Token — a Unit of Irresponsibility

What’s a token, you might ask? Those things everyone’s talking about — the expensive ones. Well, a token is a thing that shows you how irresponsible a person is. (The actual technical definition of the word “token” will follow.)

Let’s dive in.

First, let’s talk about responsibility and irresponsibility as a general subject.

I don’t think of responsibility as a “state where you ought to do something.” Not at all. Let’s look in the dictionary:

responsibility noun (plural responsibilities) the state or fact of having a duty to deal with something, or of having control over someone.

And just so we don’t miss the crucial point:

duty noun (plural duties) a moral or legal obligation.

So if you’re responsible for something, you consider that you have a duty — and you have control over it. I know that might sound terrible, but in fact it’s a great thing. If you have control over something, you can direct that something to be helpful and to survive better.

A person who is responsible for a car keeps his car in working order — clean and neat. A person who is responsible for his equipment keeps his equipment operational. An accountant who is responsible for the books keeps them reconciled.

Responsibility is not the state of being blamed for something that isn’t right. That, actually, is a good example of irresponsibility. You’re either a pro and everything’s in order, or you goofed somewhere down the line and now you’re everyone’s target.

Alright.

Now — a person who is irresponsible for an area is desperately trying to avoid it. A good manager knows something’s cooking, that someone has other fish to fry, when a guy doesn’t show up to meetings and doesn’t report on time. A person who can’t handle his car will desperately avoid going near it. If the car’s old and rusty, he’ll do everything in his power not to come near it. And if life is so “unfair” that he’s forced to use it — well, now you’ve got that well-known “numb” feeling. A guy doing something only because he’s forced to. He doesn’t want to do it, he’s not present, he doesn’t want to be present, and he just robotically goes through the motions instead of actually doing them.

Why is a person irresponsible for something? Well, that’s a sad story — and we’ll tell it next time.

A gruesome example

Let’s look at some historical examples of irresponsibility. In the good old days, when fire was a novelty and we were roaming the veldt in search of a fresh mammoth, we had a way of un-aliving others. It was rather direct — stones, or your own muscles. If Ug was a menace to you, you had a chance to go and make Ug silent forever. But this was a very direct action. You knew what you were doing, and you had to know WHAT you were getting ready to do. You had to weigh your chances and find a good rock to strike Ug down with.

Over time, however, we invented tools so we wouldn’t have to be THAT direct with Ug. Some unfortunate fellow from Rome might be un-alived by means of a spear, and over time we came up with machines to throw those spears from a great distance. The beginning of the last century “gifted” us artillery and other methods of dealing with each other. The middle of it was when we finally came up with a means of remotely destroying — if not the whole planet, then a good 90% of it.

At some point we had to take a break. People got too irresponsible about whom they decided to un-alive. Someone came to their senses, and we ended up cutting down those endless arsenals of nukes.

And yet — welcome to 2026! Modern warfare brings fresh sacrifices to the Grim Reaper’s table. We now have FPV drones that let one kill his fellow brethren from the other side of the planet. You see, there are no duels anymore — no fair fight. We’re now allowed to end someone’s life from the most distant spot you can imagine, so as not to take any responsibility for another’s life.

Well — that was a gruesome example of how this goes wrong. Now let’s look at how it plays out in the other activities of our lives.

The irresponsibility of hiring

I’ve written about this many times already. HR has been, is, and will be the most detached area of irresponsibility there is.

Some 25 years ago I could go online, find the first vacancy, do the interview, and start working two days later. 30 years ago I could just walk into a company and find a job.

What happened? We stopped taking any responsibility for our workforce. There should have been a well-organized balance, a lot of study done on how to actually employ a lot of people. Instead we let some companies run their own shows, completely detached from reality. It’s a dance of many irresponsibilities. A guy, 24 years old, full of dreams of a bright future, carrying his computer science degree, waltzes into the current job market.

It’s bloody carnage, no less.

The HR “pros” don’t want to deal with the endless stream of candidates. We don’t want to interview anyone. So we build a wall of AI-powered résumé aggregators that talk to no one and run on hidden laws nobody knows. And — thank genAI (generative artificial intelligence) — we now have AI-assisted interviews that pre-filter candidates somehow.

The candidates, on the other hand, are going in guns akimbo too. Man, the number of “résumé improvement” services that’ll drag a résumé through the ATS is incredible. And millions of résumés get posted every second — thanks, genAI.

So we literally have a situation where robots on one side are trying to stop robots on the other side from posting a résumé, while those robots are trying to get through. You see? A lot of motion, yet nobody gets hired. What does that mean? A lot of irresponsibility in one place. A good manager can always spot that place in any organization: when you see a lot of work with no ensuing production, you should know — there’s a lot of irresponsibility there.

I’m not going to go political and ask whose responsibilities those were. But know this: in most cases, these are MUTUAL irresponsibilities. There’s no justification for the fact that someone can’t find a job. Everyone is wrong here.

And now — welcome, genAI

Let’s talk about those “tokens.”

Here’s a new tech that’s supposed to make our lives easier: generative AI, now rebranded “Agentic AI.” You create so-called “agents” — electronic brains that roam on their own, executing the annoying and useless tasks in your organization. A good thing to have, I’ll grant you.

But those agents come at a price. Every word, every sign you send to this AI is a “token.” Every word you get back is also a token. So the AI companies bill you for tokens.

You’ve probably seen a bunch of memes and jokes on the internet about one company or another going bankrupt because it couldn’t pay for its tokens. How does that happen?

Well — the problem is that you can’t fix a field of irresponsibility with more irresponsibility.

If your hiring process is a mess and you don’t understand why you have to pay $25k for every hire, you should not drop an AI agent on it to “handle it” for you. It’ll end up costing you $25k plus $10k in tokens.

What you ought to do is handle it first.

Stop it.

If something doesn’t work the way it should, the first thing to do is stop it. When the “check engine” light comes on, you pull over right away — otherwise a $500 maintenance check turns into a $7,000 new-engine bill. If a process in your company yields no feasible product, you stop it.

Then you study it.

If you’re truly responsible for a part of a company — or the whole company — you should be able to go in and handle whatever’s wrong. Are your programmers spending all their time on hobby projects? Did your accountants decide to swap out the software that worked fine for ten years, and now nobody can accept a payment? Or maybe your CEO just fired the entire invoicing department and replaced it with one AI agent that has no idea how to accept a payment? I’ve seen that happen.

Study it and find out what’s wrong.

Then it’s time for a ten-bell alarm. All hands on deck. You go around and tell everyone that unless the product that used to be there starts flowing again, you’ll put a gallows up. If you’re responsible, you’ll find what someone did wrong and force him to correct it. If you’re the one who was irresponsible, you’ll have enough decency to correct it in yourself. Then you’ll write a big memorandum explaining what the hell happened and how we must never end up here again.

THEN you go and layer Agentic AI on top of a working system and make it run much better.

Yet what happens now is the opposite.

A company struggles in area A. Nobody inspects anything. We just throw agentic AI on top of the people in that area and hope they’ll figure it out. The area already doesn’t work. So now we drop AI onto it, it inspects the area, and it keeps not working — exactly the way it wasn’t working before. Except now you’ve got a lot more problems, because the AI is so diligent about finding a solution that it’ll do ANYTHING to resolve your problem. And by anything, I mean rack up a bill.

So now you’ve got a happy HR manager with an automation that replies “Thank you for your interest in the company, your résumé is under review” to every incoming email. It costs you only $2,700 a month to run, because the guy has NO idea how to automate anything. And you’ll surely recall that about a week earlier you’d asked him to handle the recruitment process so you’d have new recruits.

He’s very happy. He’s automated something in his area — an erroneous response to the thing you actually asked for, the thing he’s avoided at all costs — while distracting you with a fancy gimmick so you won’t notice that the guy actually hates his job and didn’t want to hire anyone in the first place.

My AI bill is set at $200 a month.

It’s plenty to run a LOT of work. I know what I ran and when. I can estimate my tasks, and if something doesn’t go right, I don’t just play the slot machine — I handle it first. Then, once I’ve handled the reason something went wrong, I sprinkle a little AI on top, so it CAN actually run an operational process in my company automatically.