One of the biggest risks when using generative AI is overreliance, assuming that everything it says is unquestionably true.
Take the example shown in the image. King Reccared did not sign the Third Council of Toledo as “Reccared.” This is a niche historical detail, and precisely the kind of question where a general-purpose model is likely to fail.
That’s why critical thinking is extremely important when working with AI. Models are powerful, but they can make mistakes
This is also why projects like BADDO matter. A model based in domain-specific historical knowledge can significantly reduce these errors.
So, now the question is yours:
Do you know how Reccared actually signed the III Council of Toledo?
Marina Gómez Rey
