Timnit Gebru, a researcher specializing in the ethics of artificial intelligence and particularly algorithmic bias, announced in early December the launch of an AI research institute called DAIR (Distributed Artificial Intelligence Research). DAIR aims to develop the ethics of AI and in particular applications that can have a positive impact on marginalized groups of people. This independent ecosystem-based institute would also aim to counter the pervasive influence of tech giants in AI research, development and deployment.

Timnit Gebru is a renowned researcher in artificial intelligence ethics. Born in Addis Ababa, Ethiopia in 1983, she migrated to the United States at the age of 16 and studied electrical engineering at Standford University where she earned a PhD in computer science in the Artificial Intelligence Lab in 2017. She then completed a post-doc at Microsoft in the FATE (Fairness, Accountability, Transparency and Ethics in AI) lab, dedicated to equality, transparency and ethics in AI before co-leading Google’s Ethics in AI team. She was reportedly fired last year for openly criticizing the company’s minority hiring policy and the unethical biases of AI systems.

A moratorium on certain AI applications

The topic of AI ethics is central. The issue was addressed by UNESCO in its recommendation adopted in November, but also within the European Commission and the UN. Michelle Bachelet, UN High Commissioner for Human Rights, called last September for a moratorium on certain AI applications: profiling, facial recognition, automated decision-making and machine learning technologies.

“These artificial intelligence technologies can have negative, even catastrophic, effects if used without sufficient consideration of how they affect human rights” and therefore require oversight.

According to a report published on 13 September by the UN Human Rights Council, the databases used to train AI technologies already contain biases that, if uncorrected, are transmitted to the machine and reinforce discrimination. Many examples are available on these drifts of AI, notably in recruitment, social aid, or even arrests because of badly trained facial recognition systems.

The DAIR Institute: regaining control over AI

It is in this sense, and just one year after being ousted from Google’s ethical AI team, that Timnit Gebru announced the launch of DAIR. The institute is intended as a response to the need for independent spaces where researchers from around the world can set the agenda and conduct AI research rooted in their communities and lived experiences.

According to the researcher, the inherent harm of AI is avoidable and for AI to be beneficial to all, the processes used by companies like Google, Amazon or even Meta should be challenged. In order to have better AI, researchers in the sector should be free to publish the results of their work without depending on a multinational company.

With DAIR, Timnit Gebru aims to create an environment that is independent of structures and systems that incentivise profit rather than ethics and individual well-being. She explains:

“AI needs to be brought back down to earth. It has been elevated to a superhuman level that leads us to believe that it is both inevitable and beyond our control. When AI research, development, and deployment are rooted in people and communities from the start, we can confront these harms and create a future that values equity and humanity.”

AI in the public interest

The work of independent researchers and research institutes, such as DAIR, is essential. The institute is funded by the Ford Foundation, the John D. and Catherine T. MacArthur Foundation, the Kapor Center and the Open Society Foundation. These organizations have long supported individuals and organizations, including Timnit Gebru, working to develop the field of public interest technology, an emerging field focused on harnessing the power of technology for the public good. They hope that DAIR will bring in even more public interest technologists, strengthening the movement toward inclusive and equitable technology.

Darren Walker, President of the Ford Foundation, said:

“As we continue to see, when technology is created and deployed irresponsibly, it can exacerbate systems of prejudice and discrimination with shocking speed and effect. The work of independent researchers is essential to ensure that technology has reached its potential to tackle inequality, promote inclusion and generate opportunity. Timnit Gebru’s launch and leadership of Dair will advance the field of public interest technology and ensure the movement towards ethical AI not only takes into account, but prioritizes the voices of affected communities around the world.”

John Palfrey, president of the MacArthur Foundation, said in a statement:

“Dr. Gebru is a pioneering AI researcher and advocate who has helped build a better understanding of the damage that AI technologies can potentially cause, including reinforcing racism and other forms of discrimination. To shape a more just and equitable future where AI benefits everyone, we must accelerate independent public interest research that is free of corporate constraints and focuses the expertise of people historically excluded from the AI field. MacArthur is proud to support Dr. Gebru’s bold vision for the DAIR Institute to examine and mitigate the harm caused by AI, while expanding the possibilities of AI technologies to create a more inclusive technological future.”

For Allison Scott, Ph.D, CEO of Kapor Center:

“AI researchers like Dr. Timnit Gebru have shown us that digital algorithms operating behind the scenes can be riddled with bias in design and deployment, with profound negative effects on communities of color in employment, policing, housing, health, and civic participation. We have long understood the importance of action-oriented research to drive technology advocacy. And at this pivotal moment, we’re thrilled to support and collaborate with Dr. Gebru and DAIR to increase research and advocacy on combating algorithmic bias and ensure that the promise and potential of technology is harnessed to create a more equitable future.”

As for Mark Malloch-Brown, president of the Open Society Foundations, he states:

“It’s about doing the hard work necessary to ensure that real accountability and equity are built into future AI projects, rather than as an afterthought. We need to find a way to ensure that fundamental safeguards that address accountability and fairness are at the forefront of the development of all future AI projects. We are proud to be part of a growing global movement that is holding the use of technology accountable and working to develop the collective tools and knowledge we need to verify that AI advances, rather than harms, the public interest.”

Translated from Timnit Gebru lance l’institut de recherche DAIR pour favoriser le développement éthique de l’IA