William S. Agnew
YOU?
Author Swipe
View article: Sound Check: Auditing Recent Audio Dataset Practices
Sound Check: Auditing Recent Audio Dataset Practices Open
Audio AI models are increasingly used for a broad range of applications including music and sound generation, text-to- speech (TTS), voice cloning, emotion analysis, transcription, and audio classification. However, we have little understa…
View article: Computer-vision research powers surveillance technology
Computer-vision research powers surveillance technology Open
An increasing number of scholars, policymakers and grassroots communities argue that artificial intelligence (AI) research-and computer-vision research in particular-has become the primary source for developing and powering mass surveillan…
View article: Expressing stigma and inappropriate responses prevents LLMs from safely replacing mental health providers.
Expressing stigma and inappropriate responses prevents LLMs from safely replacing mental health providers. Open
Should a large language model (LLM) be used as a therapist? In this paper, we investigate the use of LLMs to *replace* mental health providers, a use case promoted in the tech startup and research space. We conduct a mapping review of ther…
View article: Design(ing) Fictions for Collective Civic Reporting of Privacy Harms
Design(ing) Fictions for Collective Civic Reporting of Privacy Harms Open
Individually-experienced privacy harms are often difficult to demonstrate and quantify, which impedes efforts for their redress. Their effects often appear small and are inconsistently documented, and they only become more obvious when agg…
View article: The Cake that is Intelligence and Who Gets to Bake it: An AI Analogy and its Implications for Participation
The Cake that is Intelligence and Who Gets to Bake it: An AI Analogy and its Implications for Participation Open
In a widely popular analogy by Turing Award Laureate Yann LeCun, machine intelligence has been compared to cake - where unsupervised learning forms the base, supervised learning adds the icing, and reinforcement learning is the cherry on t…
View article: Sound Check: Auditing Audio Datasets
Sound Check: Auditing Audio Datasets Open
Generative audio models are rapidly advancing in both capabilities and public utilization -- several powerful generative audio models have readily available open weights, and some tech companies have released high quality generative audio …
View article: Data Defenses Against Large Language Models
Data Defenses Against Large Language Models Open
Large language models excel at performing inference over text to extract information, summarize information, or generate additional text. These inference capabilities are implicated in a variety of ethical harms spanning surveillance, labo…
View article: The Illusion of Artificial Inclusion
The Illusion of Artificial Inclusion Open
Human participants play a central role in the development of modern artificial intelligence (AI) technology, in psychological science, and in user research. Recent advances in generative AI have attracted growing interest to the possibilit…
View article: The illusion of artificial inclusion
The illusion of artificial inclusion Open
Human participants play a central role in the development of modern artificial intelligence (AI) technology, in psychological science, and in user research. Recent advances in generative AI have attracted growing interest to the possibilit…
View article: The Surveillance AI Pipeline
The Surveillance AI Pipeline Open
A rapidly growing number of voices argue that AI research, and computer vision in particular, is powering mass surveillance. Yet the direct path from computer vision research to surveillance has remained obscured and difficult to assess. H…
View article: Bound by the Bounty: Collaboratively Shaping Evaluation Processes for Queer AI Harms
Bound by the Bounty: Collaboratively Shaping Evaluation Processes for Queer AI Harms Open
Bias evaluation benchmarks and dataset and model documentation have emerged as central processes for assessing the biases and harms of artificial intelligence (AI) systems. However, these auditing processes have been criticized for their f…
View article: Representation in AI Evaluations
Representation in AI Evaluations Open
Calls for representation in artificial intelligence (AI) and machine learning (ML) are widespread, with "representation" or "representativeness" generally understood to be both an instrumentally and intrinsically beneficial quality of an A…
View article: Queer In AI: A Case Study in Community-Led Participatory AI
Queer In AI: A Case Study in Community-Led Participatory AI Open
Queerness and queer people face an uncertain future in the face of ever more widely deployed and invasive artificial intelligence (AI). These technologies have caused numerous harms to queer people, including privacy violations, censoring …
View article: Evaluating the Social Impact of Generative AI Systems in Systems and Society
Evaluating the Social Impact of Generative AI Systems in Systems and Society Open
Generative AI systems across modalities, ranging from text (including code), image, audio, and video, have broad social impacts, but there is no official standard for means of evaluating those impacts or for which impacts should be evaluat…
View article: Queer In AI: A Case Study in Community-Led Participatory AI
Queer In AI: A Case Study in Community-Led Participatory AI Open
We present Queer in AI as a case study for community-led participatory design in AI. We examine how participatory design and intersectional tenets started and shaped this community's programs over the years. We discuss different challenges…
View article: Robots Enact Malignant Stereotypes
Robots Enact Malignant Stereotypes Open
Stereotypes, bias, and discrimination have been extensively documented in\nMachine Learning (ML) methods such as Computer Vision (CV) [18, 80], Natural\nLanguage Processing (NLP) [6], or both, in the case of large image and caption\nmodels…
View article: The Values Encoded in Machine Learning Research
The Values Encoded in Machine Learning Research Open
Machine learning currently exerts an outsized influence on the world, increasingly affecting institutional practices and impacted communities. It is therefore critical that we question vague conceptions of the field as value-neutral or uni…
View article: Rebuilding Trust: Queer in AI Approach to Artificial Intelligence Risk Management
Rebuilding Trust: Queer in AI Approach to Artificial Intelligence Risk Management Open
Trustworthy artificial intelligence (AI) has become an important topic because trust in AI systems and their creators has been lost. Researchers, corporations, and governments have long and painful histories of excluding marginalized group…
View article: The Values Encoded in Machine Learning Research
The Values Encoded in Machine Learning Research Open
Machine learning currently exerts an outsized influence on the world, increasingly affecting institutional practices and impacted communities. It is therefore critical that we question vague conceptions of the field as value-neutral or uni…
View article: Documenting the English Colossal Clean Crawled Corpus.
Documenting the English Colossal Clean Crawled Corpus. Open
As language models are trained on ever more text, researchers are turning to some of the largest corpora available. Unlike most other types of datasets in NLP, large unlabeled text corpora are often presented with minimal documentation, an…
View article: Documenting Large Webtext Corpora: A Case Study on the Colossal Clean Crawled Corpus
Documenting Large Webtext Corpora: A Case Study on the Colossal Clean Crawled Corpus Open
Large language models have led to remarkable progress on many NLP tasks, and researchers are turning to ever-larger text corpora to train them. Some of the largest corpora available are made by scraping significant portions of the internet…
View article: Amodal 3D Reconstruction for Robotic Manipulation via Stability and Connectivity
Amodal 3D Reconstruction for Robotic Manipulation via Stability and Connectivity Open
Learning-based 3D object reconstruction enables single- or few-shot estimation of 3D object models. For robotics, this holds the potential to allow model-based methods to rapidly adapt to novel objects and scenes. Existing 3D reconstructio…
View article: Amodal 3D Reconstruction for Robotic Manipulation via Stability and\n Connectivity
Amodal 3D Reconstruction for Robotic Manipulation via Stability and\n Connectivity Open
Learning-based 3D object reconstruction enables single- or few-shot\nestimation of 3D object models. For robotics, this holds the potential to allow\nmodel-based methods to rapidly adapt to novel objects and scenes. Existing 3D\nreconstruc…
View article: Self-Supervised Object-Level Deep Reinforcement Learning.
Self-Supervised Object-Level Deep Reinforcement Learning. Open
Current deep reinforcement learning approaches incorporate minimal prior knowledge about the environment, limiting computational and sample efficiency. We incorporate a few object-based priors that humans are known to use: Infants divide p…
View article: Relevance-Guided Modeling of Object Dynamics for Reinforcement Learning
Relevance-Guided Modeling of Object Dynamics for Reinforcement Learning Open
Current deep reinforcement learning (RL) approaches incorporate minimal prior knowledge about the environment, limiting computational and sample efficiency. \textit{Objects} provide a succinct and causal description of the world, and many …