Pragya Singh

To be presented at #CSCW2025!
Check out: Translating Emotions to Annotations: A Participant’s Perspective of Physiological Emotion Data Collection, dl.acm.org/doi/10.1145/....

Participants self-reports are often treated as "ground-truth".

But when we asked participants to describe their process of annotating their emotions, a different story emerged.

They brought past experiences, narratives, perception towards the stimulus, experiment, and emotions into the picture.

🧵Let’s rethink how we design studies with participants as active interpreters, not just data points.

“Translating Emotions to Annotations,” explores the participant's perspective.

We found:
-Signal data and self-reports ≠ data,
-Participants aren’t passive sensors, they co-interpret signals
#cscw

Tracing the thoughts of a large language model Anthropic's latest interpretability research: a new microscope to understand Claude's internal mechanisms

Anthropic just dropped - "Tracing the thoughts of a large language model" - www.anthropic.com/news/tracing... - highly relevant for understanding LLM behind the scenes.

In the evolving world of AI it has became kind of mandatory to keep "AI in loop" to catchup with train of advancement, but it is crucial that in midst of it we don't loose what makes us human! Coz "human in loop" means dependence while "AI in loop" is agency.

Early methods for studying affective use and emotional well-being on ChatGPT An OpenAI and MIT Media Lab Research collaboration.

ChatGPT has become the ear that listens when no one else does. But seeing this made me realize how deeply people rely on AI for emotional validation—it’s alarming. It’s time we focus on building emotional literacy and openness, so people turn to each other, not machines.

openai.com/index/affect...