Pioneering Ethical AI: One Woman's Journey Away from Tech Behemoths Towards a Morally Superior AI Environment
In the world of tech, the high-profile exits of Gebru, Mitchell, and Hanna from Google have exposed systemic issues that plague the industry. These issues include diversity, transparency, and accountability in artificial intelligence (AI) development.
One organisation challenging the status quo is the Distributed AI Research Institute (DAIR). Based at https://www.dainow.org/, DAIR prioritizes research that benefits marginalized communities and addresses real-world problems. Unlike traditional research institutions, DAIR operates remotely, with teams located across the globe, reflecting its commitment to inclusivity and global perspectives.
DAIR's work significantly impacts ethical AI development by emphasizing decentralized, locally-tailored AI technologies. This approach challenges the concentration of AI power held by elite institutions and large tech companies. By addressing systemic inequalities in AI compute access, often termed "AI colonialism," DAIR advocates for distributed AI research capacity especially in under-resourced regions and academic institutions.
The institute's future vision promotes an AI ecosystem that is inclusive, equitable, and community-centered, aiming to reduce the "compute divide" by democratizing access to AI infrastructure and knowledge. DAIR critiques the current AI landscape dominated by a few corporations with proprietary infrastructure, highlighting the structural barriers that prevent diverse global participation in frontier AI research.
One of DAIR's fellows, Raesetje Sefala, is leading a project that uses satellite imagery and computer vision to study the impact of desegregation in post-apartheid South Africa. Sefala's work provides valuable insights into the long-lasting effects of apartheid and informs policy decisions aimed at social justice. Another fellow, Milagros Miceli, is researching power imbalances in the data labor industry, highlighting exploitative practices within the data annotation sector.
Dr. Alex Hanna, DAIR’s Director of Research, left Google due to limitations and lack of action within a corporate setting. Her journey from Google to DAIR demonstrates her unwavering commitment to ethical AI and a vision for dismantling systems that perpetuate inequality. Hanna's departure was preceded by the controversial exits of Timnit Gebru and Meg Mitchell from Google.
Hanna argued that Google's corporate culture was inherently biased and resistant to change, citing the "whiteness problem" within tech. Hanna's vision extends beyond identifying problems; it's about building a more just and equitable future for AI.
Other organisations working towards ethical AI include the Algorithmic Justice League (https://www.ajlunited.org/) and the Partnership on AI (https://www.partnershiponai.org/).
Looking forward, DAIR envisions an AI research future where power and resources are distributed rather than centralized, enabling academic and regional institutions worldwide to develop and govern AI autonomously. AI systems would be ethically designed with direct community involvement, increasing trust and accountability. The AI industry would confront and structurally address systemic issues such as algorithmic bias, data colonialism, and ecological impacts.
This vision complements broader AI networks that emphasize trustworthy and human-centered AI but distinctively focuses on decentralizing AI research and challenging tech industry hegemonies, thus reshaping power relations in the AI ecosystem from the ground up.
- The Distributed AI Research Institute (DAIR) emphasizes research that benefits marginalized communities and addresses real-world problems, prioritizing a decentralized approach to AI technology development to challenge the concentration of power held by elite institutions and large tech companies, as demonstrated by Dr. Alex Hanna's departure from Google.
- In the future, DAIR envisions an inclusive, equitable, and community-centered AI ecosystem, where academic and regional institutions worldwide can develop and govern AI autonomously, addressing systemic issues such as algorithmic bias, data colonialism, and ecological impacts.
- Other organizations working towards ethical AI, like the Algorithmic Justice League and the Partnership on AI, share similar visions for a future where AI events and developments are driven by insights and values prioritizing transparency, accountability, and diversity, fostering trust and equitable outcomes for all communities.
