AI products have their own failure modes

Every product can break. AI products break in some specific ways that traditional software doesn’t. Knowing those failure modes lets you design around them — and helps you spot the warning signs in your own product before they damage trust.

This topic walks through the three most common categories of AI product breakage: hallucinations, trust collapse, and cost surprises. Each has shown up across many AI products. Each is preventable with care.

Hallucinations: confident wrongness

Hallucinations are the term for AI models producing plausible-sounding but incorrect output. A made-up fact. A non-existent source. A confidently-stated answer that turns out to be wrong on closer inspection.

Hallucinations aren’t bugs in the traditional sense. They’re a pr...

Where AI products break (hallucinations, trust, cost)