In 2022, Partnership on AI (PAI) was a catalyst for positive change in the field of AI, driving improved industry practices, creating needed tools and guidance, educating policymakers and the public, and convening diverse voices to both continue our important work and identify new opportunities. As we look back on a year of impactful work, we also look forward to continuing our mission with the essential support of our Partners and other collaborators.



From Our CEO

As CEO, I’m given a privileged vantage point at PAI, one that grants me the pleasure of seeing our work in its totality. And reviewing this past year, what I find striking is the incredible breadth and depth of impact our partnership had on the field of AI.

Some of our impact was direct, like our collaboration with DeepMind to improve their data sourcing practices, two pilots using our ABOUT ML framework to operationalize transparency in the public and private sectors, and the use of PAI’s research publication recommendations in industry R&D.

Changes such as these begin by creating trustworthy tools and guidance. In 2022, PAI released 12 actionable resources for the responsible use of AI. Importantly, we did all of this in the public interest, speaking at 38 events and sharing our insights with the media, the public, and governments.

We couldn’t do this without the community of 104 Partners in 16 countries who believe in the importance and urgency of this mission. Together, we laid the groundwork for future impact by bringing together voices from across sectors and around the world.

I’m extremely proud of all that we accomplished together in 2022, only made possible through the many individual actions of our Partners, Board of Directors, and Staff. As 2023 begins, the urgency and importance of our continued work together is clear. Thank you all for the community of change you have created and continue to promote.

Rebecca Finlay
CEO, Partnership on AI
March 2, 2023

Our Year in Numbers


Changes in Practice

Our work directly resulted in improvements to transparency, documentation, and responsible practices.

Through close collaborations and the dissemination of usable guidance, PAI positively impacted industry practices in 2022. To improve the conditions of data enrichment workers, PAI worked with DeepMind to develop and implement better sourcing practices. When Meta wanted to share a new large language model, their researchers turned to PAI‘s recommendations on responsible publication norms, citing them as a key source of guidance. And with our ABOUT ML Pilots, PAI helped the UN’s Office for the Coordination of Humanitarian Affairs and a pharmacovigilance firm discover new opportunities for operationalizing transparency through documentation.

Responsible Sourcing Practices at DeepMind

Last year, PAI collaborated with DeepMind to refine and implement our recommendations for sourcing enriched data responsibly. The result was both improved practices at DeepMind and a new, practitioner-tested library of tools to improve the wellbeing of data enrichment workers.

“It’s always been a goal at DeepMind to make sure that everything we do meets a highest standards of safety and ethics, and that starts with how we build our models. That’s why we turned to Partnership on AI and the recommendations made in their white paper.”

Antonia Paterson
Science Manager, Responsible Development & Innovation Team



“The number one ESG challenge companies are grappling with is the treatment of workers. That’s why the Partnership on AI, a nonprofit focused on the responsible use of AI, studied how employers can do their part to ensure that workplace AI improves job quality rather than making it worse.”



November 16, 2022

Actionable Resources

We gave practitioners new resources for improving diversity, stakeholder engagement, and responsible AI use.

Collaboratively developing and sharing best practices is essential to advancing the AI field. In service of this, PAI provided needed guidance in 2022 for those who seek to develop and deploy AI responsibly. Last year, we released new tools for AI organizations sourcing enriched data and local newsrooms that want to adopt AI. We also offered actionable recommendations for how AI companies can make the field more inclusive, how AI conferences can foster a responsible research culture, how AI developers can engage stakeholders responsibly, and how we can all improve the impact of workplace AI.

Guidance for Improving Diversity

Last year, PAI published four recommendations for how organizations can make their AI teams more inclusive, attract and retain talent, and make good on their commitments to diversity. Drawn from in-depth interviews with minoritized AI workers and others, this guidance was designed to be ready for implementation and effective in practice.

“It’s our hope that the publication of OPT-175B will help provide a model for transparency that others in AI can follow. We need to include many voices if we want to have a responsible AI community. It’s clear that PAI’s work is bringing together that diversity of voices.”

Joelle Pineau
Managing Director
Meta AI



“[In collaboration with economist Anton Korinek, PAI is] writing a user guide for AI developers who have no background in economics to help them understand how workers might be affected by the research they are doing.”



April 19, 2022

Shared Understanding

PAI used its voice to  educate policymakers and the public about AI.

PAI shared our insights with a variety of media organizations and government agencies in 2022, informing stakeholders outside the AI community about this important technology. Last year, our work was cited by the FTC, Nature Machine Intelligence, and in the Blueprint for an AI Bill of Rights. PAI CEO Rebecca Finlay’s commentary was quoted by The White House, Fortune, Protocol, and others. PAI staffers were panelists or speakers at 38 different events around the globe, including OECD’s Digital Economy Ministerial. And after speaking on an FTC panel introduced by agency chair Lina Khan, PAI submitted a response to the FTC’s proposed rulemaking on commercial surveillance and data security practices.

PAI and the AI Bill of Rights

In its Blueprint for an AI Bill of Rights, the White House’s Office of Science and Technology Policy (OSTP) pointed to PAI’s ABOUT ML initiative as an example of how tech companies can operationalize transparency about AI systems. PAI contributed further expertise by meeting with the OSTP and attending its AI Bill of Rights launch event.

“When we wanted to improve our guidance on documentation practices, working with PAI helped us more clearly identify opportunities. By assisting us in better articulating our goals, this collaboration made imagining improvements more possible.”

Leonardo Milano
Predictive Analytics Team Lead
Centre for Humanitarian Data at UN OCHA



“The Partnership on AI did work on better understanding how AI will change the local news landscape… The takeaways here should be required reading for any journalist or media professional.”



November 4, 2022

Groundwork for Change

We brought together the international AI community to work towards positive change.

Through workshops, Partner events, and other convenings, PAI brought together voices from around the world to identify and address emergent challenges in AI. At our Quarterly Program and Steering Committee meetings, our Partners came together to identify new opportunities for improving AI practices. At workshops on ethical review practices for AI conferences, the governance of sensitive demographic data, and how to evaluate explainable AI tools, attendees shared fresh insights with our researchers. And through public comment, multiple convenings, and more than 25 individual stakeholder sessions, diverse voices helped shape PAI’s framework on responsible practices for synthetic media.

PAI and Partners in Nature Machine Intelligence

In collaboration with representatives of PAI Partners and others, PAI staff published an article in Nature Machine Intelligence advocating for coordinated support of ethics review processes in AI research. Nature Machine Intelligence’s editors later cited this article while calling for greater community discussion and consensus around publication norms.

“We are proud to be part of the multistakeholder cohort that contributed to PAI’s responsible practices for synthetic media. The challenges we face as AI develops must be discussed openly between creators and consumers. This means we must be involved in the kind of coordination that can only come through cross-sector collaboration.”

Laura Ellis
Head of Technology Forecasting

Thank You!


PAI is an independent, nonprofit 501(c)(3) organization and funded by charitable contributions from philanthropy and corporate entities. Our 990 forms from previous years can be found on our funding page.

We thank all of our Partners who form this special community for their valuable insight and support.


Philanthropic Partners

Corporate Partners

Civil Society, Academic, and Media Partners