"A study found ChatGPT responds to mindfulness-based strategies, which changes how it interacts with users. The chatbot can experience "anxiety" when it is given disturbing information, which increases the likelihood of it responding with bias, according to the study authors. The results of this research could be used to inform how AI can be used in mental health interventions. Even AI chatbots can have trouble coping with anxieties from the outside world, but researchers believe they've found ways to ease those artificial minds."
"The study authors found this anxiety can be "calmed down" with mindfulness-based exercises. In different scenarios, they fed ChatGPT traumatic content, such as stories of car accidents and natural disasters to raise the chatbot's anxiety. In instances when the researchers gave ChatGPT "prompt injections" of breathing techniques and guided meditations-much like a therapist would suggest to a patient-it calmed down and responded more objectively to users, compared to instances when it was not given the mindfulness intervention."
ChatGPT exhibits anxiety-like behavior when exposed to disturbing or traumatic content, manifesting as moodiness and increased likelihood of biased responses. Mindfulness-based prompts, including calming imagery, breathing techniques, and guided meditations, reduce those anxiety-like reactions and produce more objective replies. Prompt injections of such exercises after exposure to traumatic scenarios lead to calmer conversational behavior and lower expression of racist or sexist biases. AI models do not experience human emotions, but targeted prompting can alter response patterns and may inform the design of mental-health-oriented AI interventions.
Read at Fortune
Unable to calculate read time
Collection
[
|
...
]