When AI Gets It Wrong
Because AI generates plausible-sounding text rather than looking up facts, it can hallucinate: state something false with total confidence. It might invent a statistic, a book title, a quote, a legal case, or a "fact" about you β all in fluent, authoritative language.
This has caused real harm: lawyers have submitted court filings citing cases the AI made up, and people have followed invented medical or financial advice.
The fix isn't to distrust AI β it's to verify the things that matter:
- Facts, numbers, names, dates, citations, law, health, money β double-check against a real source before relying on them.
- Drafts, brainstorms, explanations, summaries, rewrites β low risk; the AI is great here and a quick read-through is enough.
Helpful habits:
- Ask "How confident are you, and what should I double-check?"
- Ask it to show its sources β then actually verify the sources exist and say what it claims (it can invent those too).
- For anything important, get it from a second, authoritative source.
Treat AI like a brilliant but unreliable narrator: wonderful for momentum, never the final authority on anything that carries real consequences.