2 Oct 2026 · COLLECTION

AI makes mistakes too: an experience that reminded me to check

A Russian word followed by a denial. Sharing a mistake can be a way to learn.

AI makes mistakes too. I try different models every day. Today one mixed Russian into Spanish and then denied writing that word. Poll to add: “Has this happened to you?” · “Yes” / “No” — Story · 01

A visible contradiction. First: “Temas y поведение de agente”.
Then: “The word ‘поведение’ did not appear in my previous message.” Second excerpt translated from Spanish. The error is clear; intent is not. — Story · 02

It is worth checking. Save the response. Point to the passage. Check the correction. We can benefit from AI and still review what it tells us. Question to add: “What do you usually verify?” — Story · 03

Experience caption

I try different AI models every day. Today one mixed Russian into Spanish and then denied writing that word. The screenshots let me check the contradiction. I am sharing this as a reminder: we can benefit from these tools and keep using our own judgement. What do you usually verify? Personal case from 2 October 2026. The screenshots do not establish intent or how often this happens.

My original screenshots

Original text in Spanish. The English Stories identify the translated excerpt.

Original screenshot of the AI conversation