Impactful AI: How tech leaders and social innovators are advancing AI for social good

Artificial Intelligence (AI) is a catalyst for innovation and societal change, potentially reshaping industries and addressing pressing social and environmental challenges. However, it also poses risks, including job displacement and widening wealth gaps.

The Schwab Foundation’s AI for Social Innovation initiative spotlights how social innovators use AI to impact various sectors. The following conversation — between Rowan Barnett, Director, Europe, Middle East & Africa at Google.org, Njideka U. Harry, Founder and Member of the Executive Board at Youth for Technology Foundation (YTF) and Jonathan Jackson, Founder and Chief Executive Officer of Dimagi — highlights why this dialogue is key to enabling an impactful technology roadmap. It is moderated by Daniel Nowack, Head of the Schwab Foundation's Global Alliance for Social Entrepreneurship at the World Economic Forum.

Google's investments, including the AI Opportunity Fund Europe, the Google.org AI Opportunity Fund and the Digital Futures Fund, reflect a commitment to advancing AI for social good. Through various initiatives, Google has announced over $120 million in funding on providing AI skill in recent months, with a total of $200 million previously allocated across several years.

Jonathan Jackson: AI can indeed transform how we address societal challenges like healthcare. If developed with the needs of communities in mind, it can improve access. But for that, we need to make sure that the technology is affordable, equitable, transparent and tailored to the needs of frontline workers. And on many fronts, we still have more questions than answers.

Rowan Barnett: AI can accelerate social impact while significantly reducing costs. For example, our work with MSF on Antibiogo combats antibiotic resistance, which threatens to kill 10 million by 2050. In climate action, we aided Normative in creating a carbon calculator for businesses' net-zero journey. Projects like AlphaFold, from Google DeepMind, revolutionize biological research by saving billions of hours and enable millions of researchers around the world to make discoveries (in areas including malaria vaccines, cancer treatments and enzyme design) leveraging AlphaFold’s protein structure database.

Njideka Harry: Reducing costs is important. However, with potential automation and job losses, we should also encourage job creation in high-growth industries and the AI sector.

Rowan Barnett: We’ve already seen AI's positive impact on agriculture, climate and disease detection efforts. However, we also recognize that many stakeholders lack the skills to maximize this opportunity. That is why we have launched several initiatives to address the AI skills gap — each adapted to regional needs in the US, Asia Pacific, Europe and low- and middle-income countries. The AI Opportunity Fund Europe, for example, aims to train 20,000 workers by collaborating with organizations reaching those overlooked by traditional programmes. With over 500 nonprofits expressing interest, we increased funding and announced a $15 million Google.org AI Opportunity commitment to support AI skills training in developing countries, focusing on underserved communities. This work complements Google’s broader AI skilling initiatives, such as AI Essentials for beginners, AI Startup School for entrepreneurs and Google Cloud AI programmes for businesses, as well as our Social Innovation Fund to support social entrepreneurs in Europe to leverage AI.

Rowan Barnett: Our experience has taught us the power of combining facilitated training with wraparound support and bespoke training in local languages. That way, we hope

to overcome systemic barriers, such as financial constraints, basic digital literacy or even childcare responsibilities among the target groups.

For example, over the last five years, we have funded the delivery of extensive digital skills training in countries across Europe, the Middle East and Africa, from basic digital skills to INCO’s Work in Tech programme, which supported over 30,000 people from vulnerable communities. Through extensive and tailored wraparound support, the programme increased learner completion rates from 8% up to over 48%.

Jonathan Jackson: Indeed, many partners in sub-Saharan Africa and other low and middle-income countries are curious about AI but unsure where to begin. Dimagi is committed to ensuring their voices drive progress rather than being an afterthought following high-income countries' benefits from AI. They must be integral to the progress.

Njideka Harry: Yes, human capital development is still a challenge. Youth for Technology Foundation is equipping the current and future workforce with basic to intermediate digital skills. We need African governments to partner with industry and academia to establish training for the workforce in dominant regional languages.

Njideka Harry: At the moment, nonprofits are still behind on the adoption curve. We are working hard to make AI talent democratically and equally available to different stakeholders — not just the private sector.

Rowan Barnett: Yes, this is a substantial issue. We recently surveyed over 4,000 organizations in the Google for Nonprofits programme. The survey found that four in five nonprofits see a use for generative AI in their work, but nearly half said they are not yet using the technology. Some key barriers that organizations flagged were understanding how GenAI tools can be relevant to tackling their mission and, most importantly, training. Our goal is to make this training accessible and relevant.

Jonathan Jackson: This very much mirrors our experience. Many organizations do not yet understand how much AI can do for them. For example, a lot of our partners get excited when we demonstrate role-play activities through LLMs — such as giving young married couples practice negotiating with their in-laws over birth spacing.

Rowan Barnett: The challenges are real. Technological infrastructure worldwide is unevenly distributed, and data is inconsistent and often skewed toward certain groups. We are researching the needs of vulnerable workers, but fundamentally, we believe in listening to and empowering those organizations closest to communities. Through the Digital Futures Fund, we are making sure that civil society voices are heard in the societal conversation about AI.

Jonathan Jackson: That’s a great first step, especially if it crosses cultural and language barriers and is followed up by opportunities to leverage AI. Open Chat Studio is one way to address this challenge. We know that LLMs can greatly improve and enable the work of many frontline organizations. We want to make the technology easy to use and open-source to deploy LLM-based chatbots with guardrails to improve safety and accuracy

Rowan Barnett: We recognized this issue several years ago, and that was the motivation for coming together with The Rockefeller Foundation, Canada’s International Development Resource Center (IDRC) and Germany’s GiZ FAIR Forward to launch the Lacuna Fund. Through this fund, we have support projects that fill critical data gaps, like enhancing datasets for crop yield estimation or developing resources for underrepresented languages. Sharing these learnings with the wider ecosystem is crucial. Since its inception, other funders have joined, such as the Moore Foundation, Wellcome Trust, McGovern Foundation and the Robert Wood Johnson Foundation.

Jonathan Jackson: There is still a lot to be done. We are trying to understand how LLMs can operate in low-resource languages. Simple techniques like prompting LLMs to respond in shorter, simpler sentences can already help. For example, we had good results in Chichewa in Malawi, a context where a lot of partners just assume LLMs will not work and hold back on investing.

Rowan Barnett: In 2022, Google announced an ambitious commitment to build an AI model that will support the 1,000 most spoken languages, bringing greater inclusion to billions of people in marginalized communities worldwide. We have made exciting progress toward our goal with a Universal Speech Model trained on over 400 languages, and we are seeing improved language capabilities increase the ambition of many social entrepreneurs. We recently funded a project in India to bring a personalized early education coach to every student.

Nijdeka Harry: That definitely is a great initiative. At the same time, most datasets are still largely owned by companies, stifling innovation and leading to underrepresentation. Tools like Mozilla Common Voice allow people to contribute their voice recordings in local languages to create open-source data sets. We would appreciate more of these projects going forward.

Rowan Barnett: We just co-funded Tech to the Rescue with Amazon Web Services (AWS), helping them launch their AI for Changemakers Accelerator. The programme will select over 100 nonprofit organizations and social businesses and help accelerate and amplify their impact through AI. With 1,500 tech companies involved, the tech sector is coming together and offers pro bono support and tech expertise to over 2,000 nonprofits.

Daniel Nowack: Thank you all for your insights and your important work. We are looking forward to hearing more about collaborative approaches towards AI for Good.