Maribeth Rauh
@mbrauh.bsky.social
I like to poke AI systems (sociotechnical translation and AI safety/ethics evaluation)
Researcher @aial.ie @tcddublin.bsky.social | Formerly Research Engineer @ DeepMind
Researcher @aial.ie @tcddublin.bsky.social | Formerly Research Engineer @ DeepMind
Why not be public about the definitions and methods used? Hiding so much crucial information raises major skepticism about their approach imo
October 29, 2025 at 5:47 PM
Why not be public about the definitions and methods used? Hiding so much crucial information raises major skepticism about their approach imo
What is in their "mental health taxonomy" and "emotional reliance taxonomy"? What method do they use to detect this in their data??
The results in this post are entirely predicated on those definitions which, if shoddy, could completely undermine their reported progress.
openai.com/index/streng...
The results in this post are entirely predicated on those definitions which, if shoddy, could completely undermine their reported progress.
openai.com/index/streng...
Strengthening ChatGPT’s responses in sensitive conversations
We worked with more than 170 mental health experts to help ChatGPT more reliably recognize signs of distress, respond with care, and guide people toward real-world support–reducing responses that fall...
openai.com
October 29, 2025 at 5:03 PM
What is in their "mental health taxonomy" and "emotional reliance taxonomy"? What method do they use to detect this in their data??
The results in this post are entirely predicated on those definitions which, if shoddy, could completely undermine their reported progress.
openai.com/index/streng...
The results in this post are entirely predicated on those definitions which, if shoddy, could completely undermine their reported progress.
openai.com/index/streng...