A speech-to-text slip that shows where the training came from

Always being upfront about the limitations of foundation models: I use ChatGPT’s speech-to-text a lot. In general, it transcribes about 90% correctly. I like it because I end up giving more context, since I “write more”.

However, this was the first time I saw it slip up and reveal where its training came from lol.

#GenAI is good, but you need to understand its limitations.

The original post includes an image of ChatGPT transcribing my speech as “Legendas pela comunidade Amara.org” (Portuguese for “Subtitles by the Amara.org community”).

Originally written in Portuguese and published on LinkedIn on January 17, 2025. I translated it into English with the help of AI (Claude): it let me bring my LinkedIn posts to people who don’t read Portuguese without rewriting each one from scratch. I reviewed the translation to keep my original meaning and tone.