This is an article about large language models (or LLMs). An LLM is a system trained on vast amounts of text to “learn” patterns, relationships, and structures in language, allowing it to generate content, summarize information, and assist with a wide range of tasks.

LLMs often generate information that appears convincing but is inaccurate, misleading, or completely fabricated. Because an LLM is essentially a glorified autocomplete, it predicts the most likely next words based on patterns in its training data rather than actually understanding what it’s spitting out, it can sometimes produce confident-sounding answers that are just plain wrong. We call those mistakes “hallucinations” and treat them as a normal cost of doing business.

Imagine a calculator that occasionally returned the wrong answer, or a compiler that randomly introduced bugs. We would consider these tools fundamentally broken. Yet we seem increasingly willing to tolerate factual errors from systems that are rapidly becoming involved in producing software, writing documentation, summarizing research, drafting legal documents, and generating educational material.

We’ve become accustomed to imperfection because our culture has spent decades optimizing for speed over everything else.

The normalization of LLM hallucinations is not primarily a story about LLMs. It is a story about us.

The cult of more

Modern economies reward throughput.


Companies are rewarded for growth, employees are rewarded for productivity, platforms are rewarded for engagement. Everyone is rewarded for doing more.

The result is that “faster” has gradually become synonymous with “better.”

A team that ships twice as many features is considered successful. A company that doubles output without doubling headcount is considered successful.

Whether the additional output is useful, meaningful, or even correct often becomes a secondary concern.

This logic predates AI by decades.

The industrial era taught us to think in terms of production. The internet amplified the trend. Social media accelerated it even further. LLMs are not creating a new value system, they are simply revealing the one we already had.

Creation is now effortless

For most of human history, creating something was expensive.

Writing a book was difficult, researching an article was difficult, producing software was difficult. Creating anything of value required significant effort.

As a consequence, society developed systems primarily concerned with enabling production.

Today, production is becoming cheap. A language model can generate a thousand words in seconds. It can produce documentation, reports, summaries, and code at a pace no human can match.

This changes the economics of information. When production becomes abundant, it stops being the bottleneck. Yet we continue to behave as though generating more is still the highest priority.

We celebrate the elimination of effort without asking what effort was actually accomplishing.

We “optimized” our lives the same way

Most people now approach their lives the same way organizations approach production.

Morning routines are optimized. Workflows are optimized. Relationships are optimized. Even downtime is optimized.

Activities once pursued for their own sake increasingly require justification in terms of productivity.

The language of economics has colonized everyday life.

Efficiency, optimization, output, performance, growth. These concepts are useful tools. The mistake is turning them into values.

More output does not mean more progress. Sometimes it merely creates more noise.

When everything gets louder, be quiet

In a world where everything is treated as a task, we spend our days trying to get through things. Get through the workdays, the errands, the chores, the weeks, the months. We are constantly racing toward some imagined point in the future where everything will finally be done and we will be free to enjoy ourselves.

That moment will never come.

There will always be another email, another meeting, another bill, another load of laundry, another obligation waiting around the corner. So we keep running. The free time we so desperately need, became a task in and of itself.

In the process, we forgot how to live. We forgot that a walk doesn’t have to make us healthier. A hobby doesn’t have to become a side business. A conversation doesn’t have to be productive. Not every moment needs to justify its existence.

We’ve become so obsessed with efficiency that we’ve started optimizing the life out of life.

So, maybe the answer isn’t about doing more, after all. Maybe the answer is to slow the hell down.