Key Takeaways: AI for Social Innovation Report

The social economy represents 7% of global GDP, and generative AI could add between $182 billion and $308 billion in value annually to the sector. So, how can social innovators better understand, access and deploy this technology? That’s the question at the heart of the World Economic Forum's AI for Social Innovation Initiative, which published its first report on the topic earlier this month. 

The report maps the current landscape of AI in social innovation -- its prevalence, opportunities and challenges -- drawing on a dataset of 300 social innovators and 90 AI initiatives around the world. Find a breakdown of key takeaways from the full 36-page report in our summary below: 

Geographically, social innovators in the dataset are roughly equally distributed across high and low-/middle-income countries.

Important caveat here: 300 is not a representative sample size, so we have to be careful about the conclusions we draw here. But, it does underscore the universal potential of AI to tackle a diverse range of social issues.

The US, India and the United Kingdom have the highest prevalence of social innovators deploying AI, followed by Kenya, Brazil and Nigeria.

However, the application of AI in addressing global challenges shows distinct patterns based on the economic status of the country concerned.

The three most common impact areas where AI is driving social innovation are healthcare, environment, and economic empowerment. Together, they represent about 65% of all innovators surveyed. 

If we break these impact areas down by country, some interesting patterns emerge:

• Nearly 80% of social innovators deploying AI for economic empowerment are based in low-/-middle income countries, highlighting the direct correlation between the economic priorities of a region and the focus of AI-driven social innovation within that region.

• 66% of social innovators deploying AI for environment-focused solutions are based in high income countries. 

You can see the geographical distribution patterns of all impact areas in the graphic below:

Machine learning and natural language processing are the AI capabilities most often deployed for social innovation.

The joint application of machine learning and natural language processing capabilities is the most prevalent combination because it allows organizations to analyze vast amounts of data, identify patterns, and make recommendations at high efficiency (and thus low cost.) 

Overall, approximately 10% of social innovators are deploying predictive analytics, which is most prevalent in the healthcare, economic empowerment and education domains. These innovators are using the power of predictive capabilities across areas such as predicting crop yields, monitoring water systems, supporting health diagnosis, reducing energy poverty and much more.

Women are under-represented in AI for social innovation compared to the global social enterprise space - reflecting AI’s broader gender problems

Half of all social enterprises globally are women-led -- but only 25% of social enterprises in this AI dataset are women-led. This is consistent with trends of gender in AI overall: only 22% of AI professionals globally are women, who make up less than 14% of all AI paper authors.

That being said, some regions are more balanced than others. 

• Female leadership is particularly pronounced in Oceania and Asia. 

• Africa and North America are slightly above the global average. 

• Interestingly, Europe’s share of women-led social innovators deploying AI is significantly lower than the global average, trailed only by Latin America.

When it comes to impact domains, women-led social innovators who are deploying AI are overrepresented in crisis response (75%) and far above the global average in education (40%) and healthcare (30%). They remain under-represented in areas such as security and justice (7%). Overall, AI and the technology sector remain difficult to access for women, given an inherent gender bias leading to a spiral of deterring factors.

Low and middle income countries don’t have enough ecosystem support or localized initiatives

The rapid advancement of AI in social innovation is heavily dependent on the accessibility of knowledge, data, research and resources. But the report highlights that there is a stark resource divide between high and low-/middle-income countries in that regard.

Strong AI for impact ecosystems exist in high income countries, like the USA, where nearly 43% of AI initiatives are based. Ecosystems tend to be underdeveloped in low-/middle-income countries, and a noticeable demand exists for more localized resources, tailored by geography, focus area, and economic classifications. 

Bridging AI’s trust gap and fostering collaboration are essential for its successful application in social innovation.

The report identifies a number of persistent barriers to the successful deployment of AI for social innovation. These include:

• Trust gap and systemic bias: The outputs of AI are essentially a magnification of the available data being fed into the algorithms, and the logic baked into the algorithms themselves. Biases in AI systems can deepen existing inequalities, leading to mistrust.

• Technical complexity and skills gap: The sophistication of AI demands expertise beyond many social enterprises’ capabilities. More complicated tech advancements exacerbate the existing digital divide globally.

• Resource intensity: High performance computing and data storage requirements can make AI projects unsustainable for those without substantial financial backing. Due to its energy demand, AI also has a sizable environmental footprint: according to a recent study, creating a single image through generative AI uses as much energy as a full phone charge.

• Data quality: Effective AI relies on quality data. Challenges arise when data is scarce, biased or of poor quality, leading to inaccurate AI models. Data access continues to be one of the top reported challenges for social good projects.

• Balancing social and business objectives: When leveraging communities’ assets (such as data), an inherent tension exists between revenue growth and impact. Principles like profit caps, revenue participation and profit redistribution can help reconcile the two. 

• Regulatory and ethical concerns: Regulating the use of AI to ensure its ethical application is a persistent challenge. Considerations include “privacy and surveillance, bias or discrimination, and the role of human judgment.”

• Access to AI technology: Limited access to AI technologies, especially in developing countries, hinders innovation and the adoption of AI. This does not just include access to AI solutions themselves, but also the necessary hardware, training and knowledge for assessing and implementing these systems, which leads to high “total ownership costs.” 

• Collaboration between stakeholders: Successful AI initiatives require collaboration across various stakeholders, including innovators, governments and the communities served.

If we can overcome the challenges, the benefits are vast.

AI is a powerful tool, offering a multitude of benefits that can transform the landscape of social innovation. By harnessing these advantages, organizations can amplify their impact, driving positive change on a larger scale and paving the way for a future where social innovation is more dynamic, responsive and inclusive. These include:

• Efficiency and productivity: AI’s automation capabilities are freeing up valuable time for social innovators, allowing them to focus more of strategic tasks. For example, Suki.ai (a digital assistant for doctors) automates administrative tasks with voice commands, significantly reducing the time spent on paperwork so doctors can concentrate more on patient care.

• Data-driven decision making: AI analytics are providing deep insights from large data sets, aiding in more informed decision-making. For example, MapBiomas is leveraging AI to analyze satellite imagery of the Amazon to fully map burn scars for the protection of biodiversity in the region.

• Real time responsiveness: AI enhances the ability to respond swiftly to dynamic situations, which is particularly crucial in areas like conservation. For example, The Connected Conservation Foundation in South Africa uses internet-of-things sensors with AI to create a “virtual fence” that has significantly reduced poaching with real-time alerts and animal behavior insights.

• Personalization: AI’s personalization capabilities make interventions more relevant and effective. For example, Apollo Agriculture uses AI to offer personalized farming advice to small-scale farmers in Kenya - optimizing productivity and profitability by analyzing weather patterns and soil conditions.

• Information and self-empowerment: AI is democratizing access to crucial information, and even facilitating the co-creation of public services. For example, Haqdarshak utilizes an AI platform to connect citizens with government welfare schemes in India, simplifying the application process. 

• Innovative access: Beyond making AI more efficient, AI is creating opportunities for novel ways to access essential goods and services (like education and legal advice.) For example, Barefoot Law leverages AI to provide legal assistance via mobile phones, web, radio, and community outreach programs.

• Expanding scale and reach: AI’s benefits collectively contribute to the scalability of social innovations, allowing organizations to reach broader audiences and make more significant societal impact. 

Conclusion

This report serves as a call to action for continued collaboration between technology leaders, social innovators, policy-makers, and the communities they serve in shaping an ecosystem where AI acts as a catalyst for social good.

We look forward to future reports which will delve into the frameworks for successful adoption, guidelines for responsible deployment, and strategies to positively influence the AI technology landscape. 

Want to be part of the ongoing movement to harness AI for good? Join us on the TRANSFORM Support Hub!