A new study reveals that the tech industry's ethics teams face challenges due to lack of resources and leadership support.
Researchers from Stanford University conducted interviews with AI ethics workers from major tech companies and found that these teams struggle to implement ethics initiatives in the tech industry's institutional environment.
Ethics in the Tech Field
The study was co-authored by Sanna Ali, a recent Ph.D. graduate from Stanford University, along with Stanford Assistant Professor of Communication Angèle Christin, Google researcher Andrew Smart, and Stanford W.M Keck Professor and Professor of Management Science and Engineering Riitta Katila.
Their research highlighted the difficulties faced by ethics teams in ensuring that AI products are fair, transparent, and accountable. Many companies have made public pledges to prioritize ethics in their products, but the research suggests that these promises often fall short.
Ali's interviews with AI ethics workers revealed that these teams are often under-resourced and lack leadership support to effectively address ethical concerns.
Without leadership buy-in, ethics workers have to rely on persuasive skills and interpersonal strategies to make any progress, according to Ali. As a result, their success in working with different teams varies, and they are often called for consultation too close to product launch dates, limiting their ability to enforce important ethics fixes.
The study also highlighted that the tech industry's informal and nonhierarchical structure, coupled with its emphasis on rapid product innovation, poses challenges for ethics teams in effectively integrating ethics considerations into product development processes.
As a result, the "ethics washing" phenomenon has emerged, leading to inconsistent implementation of responsible AI policies, with some products being released without proper ethics reviews.
Incentivizing Ethics Teams
To tackle these issues, Ali proposed that leadership should provide incentives to product teams to incorporate ethics considerations during their development processes.
Moreover, bureaucratic support is essential to empower ethics teams, granting them the authority to implement necessary ethics improvements before launching products.
"It's unlikely that these companies are going to change their priority of frequently releasing new products," Ali said in a statement. "But at least they could provide incentives so that ethics can be part of that conversation early on."
In addition, there is a need to incentivize product teams to collaborate more closely with ethics teams. The study noted that the current focus on speed and innovation often hampers a thorough examination of the technology's impact.
By offering rewards to teams that address product bias or tackle significant ethical concerns, companies can recognize and foster an ethical culture within their organization.
Ali advocates for establishing formal bureaucratic structures that require ethics reviews at the beginning of the product development process. This way, ethics teams won't reportedly have to convince others of the value of their work, and real institutional change may become possible.
The study was published in the Proceedings of the ACM Conference on Fairness, Accountability, and Transparency (FAccT '23).
Related Article: Computer Scientists Develop 'De-Stijl' Tool With Adobe to Help People Use Color Better in Graphic Design
