My coauthor and I are deeply concerned with the privacy threats posed by the recent developments in Artificial Intelligence and facial recognition. Governments and companies are already widely using these technologies and are well aware of their potential to identify people and their sensitive traits.
The general public, policy makers and scientists, however, seemed to be unaware of these risks. We felt that we should sound an urgent warning about the risks, show that these risks are very real, and document them with the best scientific evidence that we could produce.
Our work does not provide any value to those who may want to invade privacy. We did not develop any tools that can be used to cause harm. We only demonstrated that those already in use can reveal sexual orientation. We exposed an existing risk, but played no part in creating it.
We hope that scientists, policy makers and LGBTQ advocacy groups will work together toward the urgent, common goal of protecting the civil and human rights of LGBTQ people.
We welcome your informed feedback. Dig deeper with these relevant resources:
- Preprint text of the peer-reviewed paper, Deep neural networks are more accurate than humans at detecting sexual orientation from facial images, forthcoming in the Journal of Personality and Social Psychology.
- The summary of the findings and commonly asked questions
Statement from Paul Pfleiderer, Senior Associate Dean for Academic Affairs, Stanford Graduate School of Business
This is peer-reviewed research with pending publication in an academic journal, Journal of Personality and Social Psychology, a publication of the American Psychological Association. Publication of research findings in academic journals allows the interpretation of those findings and the research methodologies used to obtain them to be scrutinized by academics in the field and are appropriately a matter for discussion and debate.
Preamble in Stanford University Faculty Research Handbook, Adopted April 18, 1974
Stanford University's central functions of teaching, learning, research, and scholarship depend upon an atmosphere in which freedom of inquiry, thought, expression, publication, and peaceable assembly are given the fullest protection. Expression of the widest range of viewpoints should be encouraged, free from institutional orthodoxy and from internal or external coercion...