Motivated minimization and stochastic parrot

I don't like talking to people about artificial intelligence. It induces in me a mix of emotions, usually annoyance, bewilderment, slight anger, rolling eyes (not an emotion but captures it), disintegration, futility, and dullness. Every day I spend at least 5 minutes thinking about why people don't understand what is really happening with AI.

I might have an exceptional talent to find interest in minimized subjects — suddenly I realized.

Climate change — I've been working with the biggest Polish ecological organizations. I mostly ride my bike all year. When people worry I'll freeze biking through winter, I send them to Copenhagen — thousands commute by bike there in the cold, and nobody's staging a rescue. To reuse, recycle, and reduce carbon footprint is my second nature — still requires improvements. Trying to interpret why people don’t feel responsibility and keep screaming, “We can do nothing; it’s all corporations, China, India, and others’ fault,” is a hard puzzle.

Web accessibility — from a simple interest, I became a renowned expert and mentor in this hard-to-win battle, preaching, writing, teaching, and inspiring others to action. Convincing people to make the simple routine change to make web content more accessible, which most of the time does not increase workload, is sometimes a Sisyphian task.

Veganism — while I'm a non-consistent practitioner, I cook tasty vegan meals — people repeating "there's no protein in plants" or “soy lowers testosterone” drives me crazy.

And finally, AI, my last few years frustrator, and another minimized realm in the social space.

All those neglected, oftentimes serious areas, from climate change to AI, are anyway subjects I don't love getting drawn into debate with people about because of their drive to motivated minimization.

Then again, this blog and my other projects reveal an urge to deliberate and share because of a pandemic of motivated minimization.

Isn’t this contradiction wild? Well, welcome to my mind.

Motivated minimization

But what's this motivated minimization? That term appeared in my discussion with Agent Zero, my go-to tool for in-depth research. A0 coined this term. When I asked it to define it, I thought it was an established nomenclature in social psychology. It’s not!

Is A0 inventing a new term to describe a phenomenon, indicating its creativity? Oh, I will definitely get back to this soon.

Whether this term will ever appear in this context or it's just an emanation of my 2026 spring conversation with A0 living on my VPS in Helsinki, I don't care. It describes well what I see every single day, both in my conversations and in social media space.

The source nomenclature comes from a motivated reasoning coined by Ziva Kunda in 1990 ¹. People have a tendency to process information in a way that favors those parts of reality that support what they’re already motivated to believe. I am fairly certain it concerns me too 🤷🏼‍♂️.

How many times have I heard there’s no need to fix an app or a site to be accessible because users with disabilities will not use it? And here is my short, loveliest story from the accessibility field. ~7 years ago I led a plain language training for the Polish Prime Ministry authors and web editors. There was this man with a bored face; I knew it would not end well. In the middle of a workshop full of delightful revelations for most of the participants about how easy and right it is to write in an inclusive way, he exploded — “we do not write for idiots, mister.” He truly believed that it’s more appropriate to write in bureaucratic language than to write the way most people will understand. I handled the heckler, but I think he did not change his mind. Writing for people contradicted too much with his institutional background, where people race to sound smart, serious, and official. It was an example of motivated reasoning. He ignored all the plain language concept rationale, which truthfully is not too controversial and hard to acknowledge.

I have similar stories about veganism, climate change, general ecology, renewable energy, racism, poor “lazy” people, etc.

Stochastic parrot will rewire the world

And here they come. Knights of the Order of the Stochastic Parrot. While they did not even skim “On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? 🦜” paper, they have an arsenal, oh, they have. Enjoy visiting their vault in haphazard arrangement!

• “AI is a bubble like the dot-com.”
• “OpenAI is losing money.”
• “This collapses in x months — billions lost, no ROI.”
• “AI code is crap, at most at the junior level.”
• “It only predicts the next token.”
• “Scaling hit a wall.”
• “It can’t count the Rs in strawberry.”
• “Hallucinations make AI useless for serious work.”
• “AI will not replace you, only someone using AI.”
• “Luddites!”
• “Excel didn’t kill accountants.”
• “It’s not creative, just remixes training data.”
• “It’s plagiarism at an industrial scale.”
• “AI makes you dumb and intellectually atrophying.”

There’s indeed a big chance you agree with some of those quotes. Anyway, some of them have a grain of truth or even more. Those statements serve most of the time as conversation-enders. Whether those wannabe opinions are true is not the point. I have led you through the armory to illustrate that the knights weapons serve no truth or fair fight. They’re a motivated minimization: Japanese — leave-me-alone — katana cuts.

Yet, whatever weapons they use to shout at your face, this stochastic, probabilistic parrot is at this exact moment transforming the world. No advancement needed. Download one of thousands of freely available open-source LLMs and build around it.

I mean, AI itself does nothing unasked (yet²), but the alien power we have awakened is invisible to most of the people. They do not realize how transformative AI already is, do not see how it threatens the job market, or why it’s critical to understand the AI safety concerns. And this is why I don’t like to get into small talk about AI and alia; they unintentionally tend to humble those who try to see the invisible or flag the risks.

This occupies my thoughts often — why many people have this urge to minimize, bury potential threats or emerging phenomena, and crush the discussion before it starts. I will sooner or later untangle this mystery. Sometimes I grasp it, but suddenly the comprehension evaporates. Anyway, let’s dive into some of the thoughts, and they’re dull and imperfect, not a Japanese katana. But with a dull thought, I can still scratch the surface.

Why they minimize

I am far from concluding all those who minimize have bad will; many of them are smart, open minds. It’s even pretty distant from any kind of will — it emerges in the subconscious. If I had to use one simple word to describe the strongest common source of motivated minimization, it would be fear. But there are other underlying motives that shape reasoning here: human exceptionalism, believing in auto-regulation of everything, ideological spine, and normalcy bias.

In the first paragraph, my emotions and reactions — they’re undeserved. Yet, like in a solid inception, I am annoyed and rolling my eyes because of my fears of humanity not reacting to threats. So they fear, I fear, and we all source our rationale in fears. Soup of fears, hopefully vegan.

Some people know, and others feel — AI will change careers. For those for whom “LLMs are just autocomplete, stochastic parrots of sorts,” they don’t want to know or deliberate about the coming change. Anxiety overshadows their mind, and clutching at straws, they dismiss the AI aftermath on their career and accumulated expertise. Most of the people who live the life they at least accept want it to stay. Most people build their identity with their profession. So claiming AI is just a next-token predictor may at least slightly alleviate anxiety. How can such a simplistic, non-intelligent parrot replace humans? There are obvious limits to transformer architecture; it’s not a real intelligence — they repeat.

And there are others who don’t give hype an easy ride. Rational thinkers, economic trend analysts, and those who follow AI experts like Yann LeCun — they build their self-image on resisting collective excitement. “We reached the point where simple scaling will not make transformer architecture intelligent — AGI is thirty years away or can never be achieved.” Intellectually independent skeptics will not allow irrational hype to shape our social fabric. That mission allows them to construct an identity of those who are immune to hype cycles.

I scratched the surface of a few subtle reactions people express. This post got longer than I expected, so if I write a book, I promise a full chapter will cover much more, and definitely with fewer ironic flourishes.

Anyway, I don’t have good news for all those motivated to minimize. This year, I will be able to replace myself. My team, where I am the only human, can work like a small agency. I only figure out things to do, but I’m less and less necessary to actually do something, like I’ve been producing for the last 25 years. I don’t see many areas where I am smarter or more creative.