LLM Real Life | Essays by Brian Demsey

✎

What AI Hallucinations Teach Us About Ourselves

January 12, 2026 • 8 min read

We call them hallucinations—those moments when AI confidently presents fiction as fact. But as I've spent years studying these errors, I've come to see them as mirrors. They reflect not just the limitations of large language models, but something deeper about how we humans construct knowledge, memory, and meaning. Perhaps what unsettles us most about AI hallucinations isn't that machines can be wrong. It's that they're wrong in ways that feel uncomfortably familiar.

💡

The Day I Stopped Trusting AI Blindly

January 5, 2026 • 6 min read

It wasn't a catastrophic failure that changed my relationship with AI. It was a small, confident lie about a book that doesn't exist—and the realization that I almost cited it in a presentation.

💭

On Truth, Trust, and the Technology Between

December 28, 2025 • 10 min read

We've spent millennia developing systems for establishing truth—science, journalism, courts of law. Now we're training machines on the output of those systems. What happens when the student becomes the teacher?

🌎

Why I Built Hallucinations.cloud

December 15, 2025 • 7 min read

People ask what drove me to start a company focused on AI reliability. The honest answer isn't about technology or business opportunity. It's about a growing unease with how quickly we're outsourcing judgment to systems we don't fully understand.

📖

The Library of Babel Has a New Wing

December 1, 2025 • 9 min read

Borges imagined a library containing every possible book—most of them nonsense. With generative AI, we're building something similar. How do we find the signal in infinite noise?

🧠

Confidence Without Competence: AI's Most Human Trait

November 20, 2025 • 5 min read

The most unsettling thing about AI hallucinations isn't that they happen. It's that they're delivered with the same tone as accurate information. Sound familiar?

🎨

My Grandmother Never Met GPT (But She Would Have Questions)

November 8, 2025 • 8 min read

She survived an era of propaganda, learned to read between lines, and always asked "who benefits from this story?" Her questions about AI would cut right to the heart of things.

Load More Essays