Ascertain what large language model chatbots have truly learned
Develop rigorous methods to ascertain and validate the internal input–output mappings and value functions learned by large language model chatbots despite their opacity.
References
Thus, there is no easy way to know what a chatbot has truly learned.
— Technological folie à deux: Feedback Loops Between AI Chatbots and Mental Illness
(Dohnány et al., 25 Jul 2025) in Section 2, The inscrutability of large models