Automated Bullshit, Enterprise Edition

2025-12-04 — Large Language Model, Corporate Culture, Leadership — 4 min read

Human brains did not evolve to think deeply all day.

They evolved to survive while burning as little energy as possible. Thinking is expensive. Careful reasoning, holding multiple constraints in your head, exploring second- and third-order effects - all of that costs calories, time, and attention.

Whenever there is a shortcut available, the brain will try to take it.

This is not a character flaw. It is biology.

Most people, most of the time, will gravitate toward the minimum level of cognitive effort required to get by. In organisations, that usually means doing enough to look competent, useful, and busy - without necessarily producing work that is thoughtful, original, or durable.

This has always been true. What has changed is the availability of an almost perfect cognitive crutch.

Average performers are a feature, not a bug

Lets be honest: not every employee performs at the level an organisation would ideally like to see. Most people are, by definition, average. Some are below average. Some are highly capable but disengaged. And some are simply optimising for a steady paycheck and minimal friction.

At small scale, this is manageable. In early-stage startups, weak thinking is exposed quickly. There is nowhere to hide. Decisions are concrete. Code either works or it does not. Features ship or they do not.

As organisations grow, the environment changes.

Headcount increases. Communication becomes asynchronous. Output becomes more abstract. Documents, slide decks, roadmaps, and status updates replace direct problem solving. Success becomes harder to measure, and effort becomes easier to simulate.

This is the natural habitat of cognitive minimalism.

Enter the perfect shortcut

Large language models slot neatly into this dynamic.

Suddenly, the most energy-intensive part of many knowledge jobs - producing coherent text - can be outsourced to a machine. The brain no longer needs to wrestle with ideas. It only needs to provide a prompt.

That weekly update no one reads? Generated. That long Slack thread asking for alignment? Summarised. That vaguely strategic document due by end of day? Drafted in minutes.

From the outside, output increases.
From the inside, thinking decreases.

This is where the danger lies.

LLMs do not just save time. They allow people to bypass the uncomfortable, effortful phase where ideas are actually formed. The friction that used to expose shallow understanding quietly disappears.

Shallow thinking, polished output

The immediate result is not chaos. It is something far more subtle.

You get text that is fluent, confident, and well structured. Bullet points line up. Headings make sense. The tone sounds professional. The vocabulary feels reassuringly managerial.

But the content is thin.

Ideas are generic. Trade-offs are vague. Assumptions remain unexamined. Decisions are implied rather than owned. Everything sounds reasonable, and nothing is precise.

Reading this kind of output is cognitively expensive in its own way. You are forced to parse meaning that is not really there. You are no longer evaluating ideas. You are decoding noise.

Over time, this degrades the entire organisation's thinking. People respond to shallow input with equally shallow replies. Discussions become longer but less conclusive. Alignment appears to increase while clarity quietly collapses.

Camouflage for low performers

Historically, weak performers were exposed by their inability to produce coherent work.

That signal is now blurred.

With LLMs, people who struggle to reason deeply can still produce output that looks indistinguishable from that of their more capable peers. The gap between appearance and substance widens dramatically.

This makes performance harder to assess, especially in large organisations where managers already rely heavily on written artifacts to judge contribution. The people doing the hardest thinking may not look dramatically more productive than those who are mostly prompting and pasting.

Mediocrity does not just persist. It blends in.

When leaders stop thinking

The real damage starts when this pattern reaches leadership.

A manager whose primary output is LLM-generated documents can appear highly effective while contributing very little actual judgment. The slides look good. The narrative flows. The strategy sounds plausible.

But strategy without thinking is just storytelling.

When leaders outsource their reasoning, organisations lose direction without realising it. Decisions drift. Accountability dissolves. Teams sense that something is off, but cannot easily point to a single failure.

This is how organisations slowly lose their edge. Not through one bad decision, but through thousands of tiny cognitive shortcuts taken by people who no longer need to think to appear competent.

Conclusion: efficiency is not intelligence

LLMs are powerful tools. Used deliberately, they can amplify strong thinkers, reduce genuine busywork, and accelerate learning.

But they also align perfectly with the brain's natural desire to conserve energy.

If you reward volume over insight, fluency over clarity, and output over reasoning, LLMs will quietly push your organisation toward shallower thinking and lower-quality work. They will not replace your best people. They will make it harder to tell who they are.

The biggest risk is not automation. It is cognitive atrophy.

If you want LLMs to help rather than harm your business, you need systems and cultures that force thinking to stay visible. Clear ownership. Precise language. Decisions that can be traced back to real reasoning.