Nicholas Diakopoulos

Northwestern University
1 previous affiliation
4Machine learning papers tracked by Lacuna
Is this you?

Research Overview

Nicholas Diakopoulos focuses on the evaluation of digital information quality, the societal impacts of artificial intelligence, and the ethical responsibilities of the computing research community. His trajectory spans early work on assessing user-generated content and media credibility to recent investigations into how AI researchers articulate the consequences of their systems. Currently, his work addresses the broader ethical governance and accountability of machine learning research.

Research Trajectory

Nicholas Diakopoulos’s research trajectory demonstrates a sustained interest in how information quality is evaluated and how the consequences of computational systems are understood by both users and creators. His early work addressed the credibility of digital media and user-generated content, while his more recent research has shifted toward the ethical governance and societal impacts of artificial intelligence.

In 2010, Diakopoulos led research on the intersection of media credibility and user interfaces. In Modulating video credibility via visualization of quality evaluations, he developed methods for syndicating and visualizing aggregate quality assessments of informational videos. By introducing visual cues that indicated the polarity and activity levels of evaluations from media watchdogs, this work demonstrated how interface design could constructively influence casual users' perceptions of video credibility.

This focus on information quality expanded to encompass broader web ecosystems. As senior author on the 2018 study How to Assess and Rank User-Generated Content on Web, Diakopoulos guided an investigation into the features and metrics used to evaluate user-generated content (UGC). This research sought to maximize the utility of UGC for end-users and data curators, addressing the growing challenge of assessing content value in an era increasingly defined by online misinformation.

By 2021, Diakopoulos’s focus had expanded from the evaluation of content to the evaluation of computational systems themselves, specifically focusing on how the artificial intelligence community articulates the societal consequences of its research. As principal investigator on Unpacking the Expressed Consequences of AI Research in Broader Impact Statements, he led a qualitative thematic analysis of the broader impact statements introduced at the NeurIPS 2020 conference. This work categorized how researchers expressed the valence, specificity, and uncertainty of their systems' potential impacts. Methodologically, this research aligns with the broader area of Extracting Structured Themes from Qualitative Human Narratives, utilizing qualitative frameworks to understand the gap between technical design and real-world application. It also touches upon methodological rigor in qualitative studies, a concern central to the research direction of Systematic Triangulation for Reliable Qualitative Research and the area of Saturation as a Stopping Criterion for Qualitative Data, ensuring that the mapping of researcher attitudes is comprehensive and methodologically sound.

Continuing this line of inquiry, Diakopoulos participated as a contributing author in a large collaborative study in 2022, Examining Responsibility and Deliberation in AI Impact Statements and Ethics Reviews. This work analyzed the evolution of the NeurIPS ethics review process, examining the introduction of paper checklists and the shift toward explicitly addressing negative societal impacts. His involvement in this multi-author effort reflects an active engagement with the community's ongoing efforts to institutionalize ethical deliberation in machine learning research.

Current Directions

The most recent papers in the evidence indicate a current focus on the institutional and ethical mechanisms that govern artificial intelligence research. Diakopoulos is actively investigating how the machine learning community anticipates, documents, and mitigates the negative societal impacts of algorithmic systems. This involves analyzing the efficacy of ethics reviews, impact statements, and researcher checklists at major AI conferences.

A central theme of this recent work is understanding how responsibility is distributed and articulated among researchers and practitioners. This aligns with themes such as Responsibility Attribution in Distributed and Autonomous AI Systems, which explores the complex, multi-layered nature of modern AI development and the challenges of tracing failures or harms to specific interventions. Furthermore, the difficulty researchers face in accurately forecasting the downstream effects of their models resonates with work in the area of The Predictability Ceiling in Social and Physical Systems. By applying qualitative methodologies to the artifacts of AI research—such as impact statements and ethics reviews—Diakopoulos's current directions aim to bridge the gap between technical innovation and ethical accountability, ensuring that the consequences of computational systems are rigorously deliberated before deployment.

Profile updated

See something wrong? Report a mistake