Posted by Anoop Sinha, Research Director, Technology & Society, and Yossi Matias, Vice President, Google Research
The first principle of Google’s AI is to “Be socially beneficial.” As AI practitioners, we are inspired by the potential of AI technologies to transform society and our environment on a large scale and at a speed that was not possible before. From addressing the climate crisis to revolutionizing healthcare and making the digital world more accessible, our goal is to responsibly apply AI to help more people around the world. Achieving global impact requires researchers and communities to think ahead and act collectively across the AI ecosystem. This approach is called Society-Centered AI, which focuses on the needs of society as a whole while still considering the needs of individual users within the larger human experience.
Recent advancements in AI provide unprecedented capabilities at a societal level, and we now have the opportunity to systematically address these needs through collective, interdisciplinary AI research. This includes tackling challenges such as predicting hunger, forecasting diseases, and improving productivity. The potential for AI to benefit society grows every day. To understand the impact of our work in these areas, we have evaluated our research projects and initiatives. One example is the 2023 Award for Inclusion Research Program, where we selected 70 professors to support academic research that addresses the needs of historically marginalized groups globally.
Through our evaluation, we have identified emerging practices for Society-Centered AI:
1. Understanding society’s needs: It is crucial to listen to communities and partners in order to deeply understand major issues and identify priority challenges. AI has the potential to address global societal issues that significantly impact people’s lives, such as education, healthcare, and productivity. By focusing on society’s agreed-upon goals, like the United Nations’ Sustainable Development Goals, we can have a meaningful impact.
2. Collective efforts to address those needs: Bringing together stakeholders, including local communities, academic institutions, NGOs, and public-private collaborations, is essential for collectively designing, developing, implementing, and evaluating AI technologies to address societal needs.
3. Measuring success by addressing society’s needs: Measuring the effectiveness of AI solutions in addressing society’s needs is challenging but important. By collaborating with stakeholders, we can identify primary and secondary indicators of impact and optimize them.
The Society-Centered AI approach is crucial for several reasons. The following case examples demonstrate its impact across various areas, including accessibility, health, and climate:
1. Understanding the needs of individuals with non-standard speech: Project Euphonia, a Society-Centered AI initiative, allows users with non-standard speech to train personalized speech recognition models. By collaborating with non-profit organizations, Google Research developed a Universal Speech Model that significantly improved speech recognition for individuals with conditions like ALS. This led to further collaborations with other companies and organizations to make speech recognition more inclusive.
2. Focusing on society’s health needs: Google Research partnered with organizations like ARMMAN to use AI and mobile messaging to provide maternal health information to expectant and new mothers in rural India. The program resulted in improved health outcomes for mothers and infants due to the close collaboration between the community and AI technology.
3. Compounding impact from sustained efforts for crisis response: Google Research’s flood prediction efforts started with partnerships with governments and communities in India and expanded to other countries. By continually engaging with communities and refining their approach, Google Research has been able to leverage their methodologies and technology for other societal needs, such as wildfire forecasting.
To further support Society-Centered AI research, Google recently funded 18 university research proposals that exemplify this approach. These projects aim to create beneficial AI applications across the world, including monitoring peace in online conversations and developing low-cost pollution monitors to improve air quality.
In conclusion, Society-Centered AI is an approach that considers the collective needs of society while addressing the needs of individual users. By understanding society’s needs, fostering collective efforts, and measuring success based on societal impact, we can harness the transformative potential of AI for the benefit of all.
Source link
