The first time an LLM gave me exactly what I asked for, I felt uneasy. The second time, it gave me something different, better actually, an angle I hadn’t considered. The third time, it hallucinated a citation that didn’t exist.
This is when most people decide AI cannot be trusted. I decided it was teaching me something about reality I’d forgotten.
I felt that same tightness years ago, pipetting samples in a medical laboratory. Each test measured one fragment of a person’s story. A glucose level. A white blood cell count. A single data point in an ocean of biology we barely understood. We made decisions anyway. Treating the body meant accepting you’d never see the whole picture.
The work taught me to be comfortable with black boxes. To accept that you’ll never see the complete picture. To make your best guess anyway.
This post was originally published on Bronte Li’s Substack. Subscribe to receive future posts directly in your inbox.