RAISENEXUSAngela Moriah Smith / Affective scientistFund / hire

Angela Moriah Smith / Affective scientist / The RAISE Nexus

AI can increasingly produce expert work.

What happens to the inquiry while human and AI produce it together?

Since January 2025, I’ve been studying what sustained AI interaction preserves, changes, and sometimes gets wrong—before those questions had settled names.

Explore the research

A dated record of inquiry.

These questions have become
operationally consequential.

Research milestones open the evidence below. Model dates link to OpenAI’s announcements.

Explore the dated evidenceSix milestones

DEC 2025

OpenAI’s AI and Mental Health Grant Program

A proposal to turn observation into safety evaluation.

The application proposed studying which conversational patterns support agency and emotion regulation, and when safety interventions disrupt them.

Submitted December 19, 2025; not funded.

Application 0000003398 · Completed December 19, 2025

September 2026 / One example

The evaluation problem is no longer hypothetical.

Jason Locasale reported asking Astra to extend his published research program into five plausible NIH R01 proposals. The striking capability is obvious. The next research question is less obvious:

How do we know whether an AI collaborator has helped identify a better inquiry rather than generated a persuasive continuation of the existing one?

The person behind the questions

Angela
Moriah Smith

MA · they/them · San Francisco Bay Area

I notice consequential behavioral phenomena while they are still easy to dismiss as anecdotes, preserve them against premature explanation, and develop rigorous questions and tests around them.

Affective scientist · Founder of The RAISE Nexus.
Stanford BA · University of Toronto MA · Peer-reviewed affective science.