Harini Suresh

Harini Suresh

Auto-generated profile, based on publicly available information

Harini Suresh

Brown University, Providence, Rhode Island, USA

Research Focus & Metrics

Academic Career Length

10Years (2017-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
19 February 2021 in npj Digital Medicine
Verified
Journal Article
Open access
PDF file
SGSusanne Gaube
Susanne Gaube
HSHarini Suresh
Harini Suresh
MRMartina Raue
Martina Raue
AMAlexander Merritt
Alexander Merritt
SBSeth J. Berkowitz
Seth J. Berkowitz
MGMarzyeh Ghassemi
Marzyeh Ghassemi
Artificial intelligence (AI) models for decision support have been developed for clinical settings such as radiology, but little work evaluates the potential impact of such systems. In this study, physicians received chest X-rays and diagnostic advice, some of which was inaccurate, and were asked to evaluate advice quality and make diagnoses. All advice was generated by human experts, but some was labeled as coming from an AI system. As a group, radiologists rated advice as lower quality when it appeared to come from an AI system; physicians with less task-expertise did not. Diagnostic accuracy was significantly worse when participants received inaccurate advice, regardless of the purported source. This work raises important considerations for how advice, AI and non-AI, should be deployed in clinical environments.

3 June 2024
Verified
Conference Paper
Open access
HSHarini Suresh
Harini Suresh
ETEmily Tseng
Emily Tseng
MYMeg Young
Meg Young
MGMary Gray
Mary Gray
EPEmma Pierson
Emma Pierson
KLKaren Levy
Karen Levy
Growing interest and investment in the capabilities of foundation models has positioned such systems to impact a wide array of services, from banking to healthcare. Alongside these opportunities is the risk that these systems reify existing power imbalances and cause disproportionate harm to historically marginalized groups. The larger scale and domain-agnostic manner in which these models operate further heightens the stakes: any errors or harms are liable to reoccur across use cases. In AI & ML more broadly, participatory approaches hold promise to lend agency and decision-making power to marginalized stakeholders, leading to systems that better benefit justice through equitable and distributed governance. But existing approaches in participatory AI/ML are typically grounded in a specific application and set of relevant stakeholders, and it is not straightforward how to apply these lessons to the context of foundation models. Our paper aims to fill this gap. First, we examine existing attempts at incorporating participation into foundation models. We highlight the tension between participation and scale, demonstrating that it is intractable for impacted communities to meaningfully shape a foundation model that is intended to be universally applicable. In response, we develop a blueprint for participatory foundation models that identifies more local, application-oriented opportunities for meaningful participation. In addition to the “foundation” layer, our framework proposes the “subfloor” layer, in which stakeholders develop shared technical infrastructure, norms and governance for a grounded domain such as clinical care, journalism, or finance, and the “surface” (or application) layer, in which affected communities shape the use of a foundation model for a specific downstream task. The intermediate “subfloor” layer scopes the range of potential harms to consider, and affords communities more concrete avenues for deliberation and intervention. At the same time, it avoids duplicative effort by scaling input across relevant use cases. Through three case studies in clinical care, financial services, and journalism, we illustrate how this multi-layer model can create more meaningful opportunities for participation than solely intervening at the foundation layer.


23 January 2025 in NEJM AI
Verified
Journal Article
Closed access
EPEmma Pierson
Emma Pierson
DSDivya Shanmugam
Divya Shanmugam
RMRajiv Movva
Rajiv Movva
JKJon Kleinberg
Jon Kleinberg
MAMonica Agrawal
Monica Agrawal
KVKeyon Vafa
Keyon Vafa

24 November 2020 in arXiv
Verified
Preprint
Open access
PDF file
ADAlexander D'Amour
Alexander D'Amour
KHKatherine Heller
Katherine Heller
DMDan Moldovan
Dan Moldovan
BABen Adlam
Ben Adlam
BABabak Alipanahi
Babak Alipanahi
DSD. Sculley
D. Sculley
ML models often exhibit unexpectedly poor behavior when they are deployed in real-world domains. We identify underspecification as a key reason for these failures. An ML pipeline is underspecified when it can return many predictors with equivalently strong held-out performance in the training domain. Underspecification is common in modern ML pipelines, such as those based on deep learning. Predictors returned by underspecified pipelines are often treated as equivalent based on their training domain performance, but we show here that such predictors can behave very differently in deployment domains. This ambiguity can lead to instability and poor model behavior in practice, and is a distinct failure mode from previously identified issues arising from structural mismatch between training and deployment domains. We show that this problem appears in a wide variety of practical ML pipelines, using examples from computer vision, medical imaging, natural language processing, clinical risk prediction based on electronic health records, and medical genomics. Our results show the need to explicitly account for underspecification in modeling pipelines that are intended for real-world deployment in any domain.

25 April 2025
Verified
Conference Paper
Open access
KFK. J. Kevin Feng
K. J. Kevin Feng
RPRock Yuren Pang
Rock Yuren Pang
TKTzu-Sheng Kuo
Tzu-Sheng Kuo
AWAmy Winecoff
Amy Winecoff
ETEmily Tseng
Emily Tseng
AZAmy X. Zhang
Amy X. Zhang