Democratizing AI for the Public Good

Key Concepts and Recommendations
October 3, 2024

AI technology can be a powerful tool to advance the Sustainable Development Goals (SDGs) if designed, applied, and governed responsibly. Research shows that the AI infrastructure stack – from hardware to cloud infrastructure, data, models, and the application layer -is highly skewed towards closed research and less transparency despite increased “open-washing.”

This policy brief, written for the Brasilian G20 Presidency, outlines how stakeholders can address the layers of AI democratization (use, development, benefit, governance). Strategies for democratizing AI can also help deal with adverse results of AI development: The report highlights how AI developments result in economic, social, and political challenges: limited innovation and quality, risks of bias and disinformation, digital extremism, and a lack of democratic control and digital sovereignty. Specifically, the policy brief considers open-source AI as a digital public good that can be instrumental in progressing toward AI democratization and accelerating the attainment of the SDGs.

The brief is co-authored by Lea Gimpel (Digital Public Goods Alliance, Norway); Daniel Brumund (GIZ “FAIR Forward – Artificial Intelligence for All”, Germany); Alek Tarkowski (Open Future Europe, Poland); Maximilian Gahntz (Mozilla Foundation, Germany); Urvashi Aneja (Digital Futures Lab, India); Vukosi Marivate (University of Pretoria, South Africa); and Anita Gurumurthy (IT for Change, India).

Policy recommendations focus on four main areas:

  1. Provide public AI infrastructure and develop a shared strategy for developing public generative AI models
  2. Promote the creation, curation, and governance of trusted high-value open datasets
  3. Develop industry standards for the safe disclosure of model weights and related AI assets
  4. Harmonize AI governance for the public good

The final part of the report provides a scenario of outcomes addressing the widespread debate about the open-source foundational AI model’s risks and benefits.

“By developing policy instruments that promote AI systems and their components as DPGs, the world community can counter an increasingly present AI monoculture and support global majority countries in seizing AI’s potential for economic prosperity and people’s and planetary well-being.”

The T20 is a G20 engagement group that brings together think tanks and research centers from G20 members, guest countries, and organizations. Task Force 5 (TF05) on Inclusive digital transformation centers its efforts on elaborating recommendations to leverage digital innovations to advance the SDGs of the 2030 Agenda while ensuring inclusivity and ethical considerations.

Read the brief

Alek Tarkowski
with: Lea Gimpel (Digital Public Goods Alliance (DPGA)), Daniel Brumund (GIZ “FAIR Forward – Artificial Intelligence for All”), Maximilian Gahntz (Mozilla), Urvashi Aneja (Digital Futures Lab), Vukosi Marivate (University of Pretoria), Anita Gurumurthy (IT for Change)
download as PDF:
keep up to date
and subscribe
to our newsletter
Subscribe