Research Foundations

Research Papers

This report catalogs 60+ publications, institutional pages, and framework documents across five research domains.

Sources span peer-reviewed journals (Nature, JAMA, NeurIPS, ACL), preprint archives (arXiv), and official institutional pages from Stanford, Harvard, Google, Microsoft, Meta, Anthropic, OpenAI, and Georgia Tech.

Crisis Detection & Mental Health NLP Human-in-the-Loop AI Governance Conversational Analysis & Pastoral Language Behavioral Engagement Modeling AI Safety & Model Alignment

Key Institutions

Stanford HAI

Human-Centered AI Institute

Stanford CRFM

Center for Research on Foundation Models

Harvard Berkman Klein Center

Internet & Society

Harvard T.H. Chan School

Onnela Lab — Digital Phenotyping

University of Washington

Tim Althoff's Research Group

1. Crisis Detection & Mental Health NLP

Smartphone-Based Conversational Agents and Responses to Questions About Mental Health

Adam S. Miner, Arnold Milstein, Stephen Schueller, et al.

JAMA Internal Medicine, 2016

Landmark audit of how Siri, Google Now, Cortana, and S Voice respond to crisis statements.

Natural language processing system for rapid detection and intervention of mental health crisis chat messages (CMD-1)

Akshay Swaminathan, Ivan Lopez, Jonathan H. Chen, et al.

npj Digital Medicine, 2023

NLP system detecting suicidal/homicidal ideation with AUC of 0.98, reducing triage time from ~9 hours to under 10 minutes.

Artificial Intelligence and Suicide Prevention: A Systematic Review

Rebecca A. Bernert, Amanda M. Hilberg, Ruth Melia, et al.

International Journal of Environmental Research and Public Health, 2020

Systematic review of 87 AI/ML studies on suicidal behavior classification, reporting >90% accuracy.

Key Considerations for Incorporating Conversational AI in Psychotherapy

Adam S. Miner, Nigam Shah, Kim D. Bullock, et al.

Frontiers in Psychiatry, 2019

Using NLP to Detect Mental Health Crises

Stanford HAI

Stanford HAI News

Large Language Models in Mental Health Care: A Scoping Review

Yining Hua, Fenglin Liu, Kailai Yang, et al.

arXiv, 2024

First comprehensive review of LLMs in mental health care, examining 34 studies. Harvard T.H. Chan School of Public Health.

A scoping review of large language models for generative tasks in mental health care

Yining Hua, Hongbin Na, Zehan Li, et al.

npj Digital Medicine, 2025

Natural language processing applied to mental illness detection: a narrative review

Tianlin Zhang, Annika M. Schoene, Shaoxiong Ji, Sophia Ananiadou

npj Digital Medicine, 2022

Comprehensive review of 399 NLP studies for mental illness detection.

Methods in predictive techniques for mental health status on social media: a critical review

Stevie Chancellor, Munmun De Choudhury

npj Digital Medicine, 2020

Review of 75 studies (2013-2018) on predicting mental health from social media. 406+ citations.

Large-scale Analysis of Counseling Conversations: An Application of NLP to Mental Health

Tim Althoff, Kevin Clark, Jure Leskovec

TACL, 2016 — Best Paper Award at WWW

Analysis of 660,000 text messages from 15,000 Crisis Text Line sessions.

Towards Facilitating Empathic Conversations in Online Mental Health Support

Ashish Sharma, Inna W. Lin, Adam S. Miner, et al.

WWW 2021 — Best Paper Award

A Computational Approach to Understanding Empathy Expressed in Text-Based Mental Health Support

Ashish Sharma, Adam S. Miner, David C. Atkins, Tim Althoff

EMNLP 2020

Human–AI collaboration enables more empathic conversations in text-based peer-to-peer mental health support

Ashish Sharma, Inna W. Lin, Adam S. Miner, et al.

Nature Medicine, 2023

RCT with 300 peer supporters showing a 19.6% increase in conversational empathy via AI-in-the-loop.

Predicting Depression via Social Media

Munmun De Choudhury, Michael Gamon, Scott Counts, Eric Horvitz

ICWSM 2013

Foundational work from Georgia Tech on social media-based depression prediction.

2. Human-in-the-Loop AI Governance

Guidelines for Human-AI Interaction

Saleema Amershi, Dan Weld, Mihaela Vorvoreanu, et al.

CHI 2019 — Award-winning

18 design guidelines for human-AI interaction, validated with 49 design practitioners across 20 AI products.

Datasheets for Datasets

Timnit Gebru, Jamie Morgenstern, Briana Vecchione, et al.

Communications of the ACM, 2021

Co-Designing Checklists to Understand Organizational Challenges around Fairness in AI

Michael Madaio, Luke Stark, Hanna Wallach, Jennifer Wortman Vaughan

CHI 2020 — Best Paper Award

Microsoft HAX Toolkit

Microsoft Research

Microsoft

Microsoft Responsible AI Standard v2

Microsoft

Government by Algorithm: Artificial Intelligence in Federal Administrative Agencies

David Freeman Engstrom, Daniel E. Ho, Catherine M. Sharkey, et al.

Stanford HAI, 2020

The most comprehensive study of AI use in U.S. federal agencies at the time.

Stanford HAI AI Index Report 2025

Stanford HAI

Principled Artificial Intelligence: Mapping Consensus in Ethical and Rights-Based Approaches

Jessica Fjeld, Nele Achten, Hannah Hilligoss, et al.

Harvard Berkman Klein Center, 2020

Compared 36 AI principles documents, identifying eight consensus themes across global AI governance frameworks.

Model Cards for Model Reporting

Margaret Mitchell, Simone Wu, Andrew Zaldivar, et al.

FAT* 2019

Introduced model cards — now an industry standard adopted across platforms like Hugging Face.

Google Responsible AI Practices

Google

Google AI

People + AI Research (PAIR) Guidebook

Google PAIR

Google

3. Conversational Analysis & Pastoral Language

A Simplest Systematics for the Organization of Turn-Taking for Conversation

Harvey Sacks, Emanuel A. Schegloff, Gail Jefferson

Language, 1974

The foundational paper of conversation analysis (~10,000+ citations).

The Preference for Self-Correction in the Organization of Repair in Conversation

Emanuel A. Schegloff, Gail Jefferson, Harvey Sacks

Language, 1977

Lectures on Conversation, Volumes I and II

Harvey Sacks (edited by Gail Jefferson)

Wiley-Blackwell, 1992

The foundational text of conversation analysis.

Sequence Organization in Interaction: A Primer in Conversation Analysis

Emanuel A. Schegloff

Cambridge University Press, 2007

Towards Empathetic Open-domain Conversation Models: A New Benchmark and Dataset

Hannah Rashkin, Eric Michael Smith, Margaret Li, Y-Lan Boureau

ACL 2019

Introduced the EmpatheticDialogues dataset (~25k conversations).

It Sounds Like...: A Natural Language Processing Approach to Detecting Counselor Reflections in Motivational Interviewing

Michael Tanana, Kevin A. Hallgren, Zac E. Imel, et al.

Journal of Counseling Psychology, 2016

Achieved 93% recall and 90% specificity for detecting counselor reflections.

Natural Language Processing for Mental Health Interventions: A Systematic Review

Matteo Malgaroli et al.

Translational Psychiatry, 2023

Review of 102 NLP studies applied to mental health interventions.

A Computational Framework for Behavioral Assessment of LLM Therapists

Yu Ying Chiu, Ashish Sharma, Inna Wanyin Lin, Tim Althoff

IJCAI 2024

Therapy Talk and Therapeutic Conversations: The Formation of Pastoral Counselors

Peter L. VanKatwyk

Journal of Pastoral Care & Counseling, 2006

Communicating in Chaplaincy: Fundamentals and Case Studies

Stefan Gärtner

Journal of Health and Social Care Chaplaincy, 2024

Three-level communication model for chaplaincy.

An Empirical Study on the Nature of the Verbal Responses of Humanist Chaplains

Various

Religions, 2021 (MDPI)

4. Behavioral Engagement Modeling

Deep Neural Networks for YouTube Recommendations

Paul Covington, Jay Adams, Emre Sargin

RecSys 2016

Foundational paper on YouTube's two-stage deep learning recommendation system.

Wide & Deep Learning for Recommender Systems

Heng-Tze Cheng, Levent Koc, et al.

DLRS Workshop, 2016

Recommending What Video to Watch Next: A Multitask Ranking System

Zhe Zhao, Lichan Hong, Li Wei, et al.

RecSys 2019

YouTube's multi-objective ranking system using Multi-gate Mixture-of-Experts (MMoE).

Retentive Relevance: Capturing Long-Term User Value in Recommendation Systems

Saeideh Bakhshi et al. (Meta)

arXiv, 2025

Content-level measure outperforming engagement signals in predicting next-day retention. Deployed on Facebook.

Deep Learning Recommendation Model (DLRM)

Maxim Naumov, Dheevatsa Mudigere, et al. (Meta)

arXiv, 2019

Meta's foundational open-source recommendation model architecture.

Harnessing Smartphone-Based Digital Phenotyping to Enhance Behavioral and Mental Health

Jukka-Pekka Onnela, Scott L. Rauch

Neuropsychopharmacology, 2016

Foundational paper defining digital phenotyping.

Relapse Prediction in Schizophrenia Through Digital Phenotyping

Ian Barnett, John Torous, Patrick Staples, et al.

Neuropsychopharmacology, 2018

Key paper on personal baseline deviation detection using the Beiwe app.

Anomaly Detection to Predict Relapse Risk in Schizophrenia

Various

Translational Psychiatry, 2020

89% sensitivity and 75% specificity for predicting relapse via smartphone-based behavioral anomaly detection.

Beiwe: A Data Collection Platform for High-Throughput Digital Phenotyping

Jukka-Pekka Onnela et al.

JOSS, 2021

5. AI Safety & Model Alignment

Constitutional AI: Harmlessness from AI Feedback

Yuntao Bai, Saurav Kadavath, Sandipan Kundu, Amanda Askell, et al.

Anthropic, 2022

Introduced Constitutional AI (CAI), training harmless AI via principles using RLAIF.

Training a Helpful and Harmless Assistant with RLHF

Yuntao Bai, Andy Jones, et al. (Anthropic)

arXiv, 2022

Collective Constitutional AI: Aligning a Language Model with Public Input

Deep Ganguli et al. (Anthropic)

Anthropic, 2023

Used input from ~1,000 Americans via Polis to draft a publicly sourced constitution.

Claude's Constitution

Anthropic

Holistic Evaluation of Language Models (HELM)

Percy Liang, Rishi Bommasani, Tony Lee, et al.

TMLR, 2023

Evaluates 30 LMs across 7 metrics and 16 core scenarios.

On the Opportunities and Risks of Foundation Models

Rishi Bommasani, Drew A. Hudson, et al. (100+ Stanford researchers)

Stanford CRFM, 2021

200+ page report defining and analyzing foundation models.

Training Language Models to Follow Instructions with Human Feedback (InstructGPT)

Long Ouyang, Jeff Wu, Xu Jiang, et al. (OpenAI)

NeurIPS 2022

The landmark RLHF paper establishing the three-stage RLHF process as industry standard.

Learning to Summarize from Human Feedback

Nisan Stiennon, Long Ouyang, Jeff Wu, et al.

NeurIPS 2020

Foundational precursor to InstructGPT.

Weak-to-Strong Generalization

Collin Burns, Jan Leike, Leopold Aschenbrenner, et al. (OpenAI Superalignment)

OpenAI, 2023

Studies whether weak models can supervise strong ones.

Training Value-Aligned Reinforcement Learning Agents Using a Normative Prior

Md Sultan Al Nahian, Spencer Frazier, Brent Harrison, Mark Riedl

arXiv, 2021

Responsible AI Agents

Deven R. Desai and Mark Riedl

arXiv, 2025

Framework for responsible AI agent deployment.