# Alice Xiang

Alice Xiang is a lawyer, statistician, and Global Head of AI Governance at Sony Group Corporation, known for her work on AI ethics, fairness, and ethical data curation. She was named by Nature as one of ten scientists to watch in 2026.

Alice Xiang is a lawyer, statistician, and the Global Head of AI Governance and Lead Research Scientist at Sony Group Corporation. Her work focuses on the intersection of law, statistics, and [artificial intelligence](https://www.wikiprompt.org/wiki/artificial-intelligence), particularly on issues of fairness, transparency, and accountability in AI systems. In 2026, Nature named her one of ten scientists to watch, recognizing her contributions to developing ethical benchmarks for AI data. She is also known for leading the creation of the Fair Human-Centric Image Benchmark (FHIBE), a dataset of over 10,000 images collected with explicit attention to diversity, bias mitigation, and consent.

Xiang's career spans academia, industry, and non-profit research. Before joining Sony, she worked at the Partnership on AI, where she headed fairness, transparency, and accountability research. Her academic training includes degrees from Harvard University, the University of Oxford, and Yale Law School, giving her a multidisciplinary perspective that she applies to the governance of emerging technologies. She has authored influential papers and op-eds on topics such as skin tone bias, ethical data curation, and the legal and technical challenges of algorithmic fairness.

## Education and Early Career

Xiang earned a Juris Doctor from Yale Law School, a master's degree in development economics from the [University of Oxford](https://www.wikiprompt.org/wiki/oxford-university), and both a bachelor's degree in economics and a master's degree in statistics from Harvard University. This combination of legal and quantitative training is central to her approach to AI ethics, allowing her to bridge the gap between technical model development and regulatory frameworks.

Early in her career, Xiang worked at the Partnership on AI, a non-profit organization founded by major tech companies including [OpenAI](https://www.wikiprompt.org/wiki/openai), [Google DeepMind](https://www.wikiprompt.org/wiki/google-deepmind), [Apple](https://www.wikiprompt.org/wiki/apple), and [Amazon Web Services](https://www.wikiprompt.org/wiki/amazon-web-services). At the Partnership on AI, she served as the head of fairness, transparency, and accountability research, where she led projects examining how AI systems can be made more equitable and how their decisions can be explained to users. She also held a visiting scholar position at Tsinghua University in Beijing, further broadening her international perspective on AI governance.

## Leadership at Sony

Xiang joined Sony Group Corporation as Global Head of AI Governance and Lead Research Scientist. In this role, she oversees the company's strategies for ensuring that its AI products and research adhere to ethical principles. Sony's AI initiatives span a wide range of applications, from consumer electronics to entertainment and gaming, and Xiang's work involves developing governance frameworks that can be applied across these diverse domains.

Under her leadership, Sony has emphasized the importance of ethical data collection and curation. Xiang has argued that the quality and fairness of AI systems depend heavily on the data used to train them, and she has advocated for industry-wide standards that prioritize consent, diversity, and intellectual property protection. Her work at Sony has positioned the company as a leader in responsible AI development, particularly in the context of [generative AI](https://www.wikiprompt.org/wiki/generative-ai) and [large language models](https://www.wikiprompt.org/wiki/large-language-model).

## Fair Human-Centric Image Benchmark (FHIBE)

One of Xiang's most significant contributions is the development of the Fair Human-Centric Image Benchmark (FHIBE), a dataset of more than 10,000 images of humans. The dataset was created with the explicit goal of reflecting diversity, mitigating bias, protecting intellectual-property rights, and including consent from the individuals photographed. This is a departure from many existing image datasets, which are often scraped from the internet without explicit permission and may underrepresent certain demographic groups.

FHIBE was published in a paper in Nature in November 2025, co-authored with Jerone T. A. Andrews, Rebecca L. Bourke, and others. The dataset is intended to serve as a benchmark for evaluating AI models' performance across different demographic groups, helping researchers identify and correct biases in [machine learning](https://www.wikiprompt.org/wiki/machine-learning) systems. In 2026, FHIBE was recognized as one of Fast Company's World Changing Ideas, highlighting its potential to reshape how AI training data is sourced and used.

## Research on Fairness and Transparency

Xiang has published extensively on the challenges of ensuring fairness in AI. Her work often addresses the tension between technical approaches to bias mitigation and the legal and social contexts in which AI systems operate. In her paper "Reconciling legal and technical approaches to algorithmic bias," published in the Tennessee Law Review in 2020, she explored how legal standards for discrimination can inform the design of fairer algorithms, and vice versa.

Another notable contribution is her 2021 paper "What We Can't Measure, We Can't Understand: Challenges to Demographic Data Procurement in the Pursuit of Fairness," co-authored with McKane Andrus, Elena Spitzer, and Jeffrey Brown. The paper examines the practical difficulties of collecting demographic data needed to audit AI systems for bias, including privacy concerns and the risk of misclassification. This work has been influential in shaping debates about how to balance fairness and privacy in AI.

Xiang has also investigated the concept of uncertainty as a form of transparency. In a 2021 paper presented at the AAAI/ACM Conference on AI, Ethics, and Society, she and her colleagues argued that AI systems should communicate their uncertainty to users, allowing for more informed decision-making. This perspective challenges the common assumption that AI outputs should always be presented with high confidence.

## Ethical Harms of Speech Generators

In 2024, Xiang co-authored a paper titled "Not My Voice! A Taxonomy of Ethical and Safety Harms of Speech Generators" with Wiebke (Toussaint) Hutiri and Orestis Papakyriakopoulos. The paper, presented at a conference organized by the Association for Computing Machinery, provides a comprehensive framework for categorizing the harms that can arise from AI-generated speech. These include identity theft, impersonation, and the spread of misinformation, as well as less obvious harms such as the erosion of trust in audio evidence.

The taxonomy is intended to help developers and policymakers anticipate and mitigate the risks associated with speech generation technologies, which have become increasingly sophisticated with the advent of [deep learning](https://www.wikiprompt.org/wiki/deep-learning) and [neural networks](https://www.wikiprompt.org/wiki/neural-network). Xiang's work in this area reflects her broader concern with the societal implications of AI, particularly as generative models become more capable of producing realistic content.

## Recognition and Awards

Xiang's contributions have been widely recognized. In 2021, she was included in the 100 Brilliant Women in AI Ethics list by Women in AI Ethics. In 2025, she received the Privacy Papers for Policymakers Award from the Future of Privacy Forum for her essay "Mirror, Mirror, on the Wall, Who's the Fairest of Them All?," which was published in the journal Daedalus in 2024. The essay examines how computer vision systems can mis-see individuals, particularly those from marginalized groups, and argues for a more nuanced understanding of fairness that accounts for both privacy and visibility.

In 2026, she was named one of the Top 100 Women in AI by AI Magazine, and Nature included her in its list of ten scientists to watch. These honors reflect her growing influence in the field of AI ethics, where she is seen as a bridge between technical research and policy development.

## Selected Works and Publications

Xiang has authored or co-authored numerous influential papers. Among her most cited works is "Explainable machine learning in deployment," published in 2020, which examines the practical challenges of implementing explainability techniques in real-world systems. The paper, co-authored with Umang Bhatt and others, discusses the gap between academic research on interpretability and the needs of practitioners.

Her publication record also includes work on the legal dimensions of AI, such as "Being 'seen' versus 'mis-seen': Tensions between privacy and fairness in computer vision," published in the Harvard Journal of Law & Technology in 2022. This paper explores how privacy protections can conflict with efforts to ensure fairness, particularly in surveillance and facial recognition technologies.

In addition to academic papers, Xiang has written op-eds for Fortune and Time magazine, where she has communicated her research on skin tone bias and ethical data curation to a broader audience. Her ability to translate complex technical issues into accessible language has made her a sought-after commentator on AI ethics.

## Impact and Legacy

Xiang's work is part of a broader movement to make AI more accountable to the people it affects. By focusing on the entire lifecycle of AI systems - from data collection to deployment - she has helped shift the conversation from purely technical performance to questions of social responsibility. Her leadership at Sony, combined with her academic contributions, positions her as a key figure in the ongoing effort to develop AI that is both innovative and ethical.

As AI continues to permeate every aspect of society, the frameworks Xiang has developed for ethical data curation and fairness assessment are likely to become increasingly important. Her emphasis on consent and diversity in data collection, in particular, offers a model for how companies can build AI systems that respect individual rights while still achieving high performance. Through her research, advocacy, and leadership, Xiang is helping to define what responsible AI looks like in the 21st century.

---
Source: https://www.wikiprompt.org/wiki/alice-xiang
License: CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/)
Last updated: 2026-09-09T01:57:38.415793+00:00