Technology often asks us to hold two seemingly competing experiences together at once. Think of the love-hate relationship you have with your phone and social media. If you watched the Oppenheimer movie, you saw a prime example of this. While nuclear energy is used in medicine to treat diseases like cancer, it was also used to build and deploy the atomic bomb.
Now with AI, we are at the next trajectory in our relationship with technology. News reports on AI’s threat to human agency and life abound. AI has perpetuated harms and deepened existing negative outcomes for communities, especially for those already facing social inequities. Examples include predictive policing, facial recognition technology, and racially profiling black and brown people.
In April 2022, the AP obtained research from a Carnegie Mellon University team showing an algorithm that screens for child neglect in Allegheny County, Pennsylvania, displayed a pattern of flagging a disproportionate number of Black children for a “mandatory” investigation.
Algorithms used to predict whether women could have vaginal births with a second child after a c-section birth wrongly led to greater c-sections in African American and Hispanic women. In this case, algorithms used existing race based medical data, and biased outcomes for non-white women, pushing the maternal health pendulum in a further negative direction. These examples can lead people to believe that AI is biased and not worth engaging with.
Alongside actively engaging with AI’s risk to humanity and our planet, we will need to hone dialectical thinking and actions to engage with its positive transformative potential. This will mean adopting a both/and approach. In a study by Miron-Spektor et al., 2018, they found that, “Leaders who are able to hold opposing perspectives simultaneously are more likely to see the potential for creative solutions.”
When we think of technology as a catalyst for change, the most popular headlines tend to skew on the risks of tech and the impact of big tech companies. What we hear less about is the opportunity for more and more people, not just technologists, to create solutions and products that shift who can solve some of the most pressing issues of our time, when, where and how. What if we think of AI as a partner in expanding opportunity, solutions and access to addressing some of the world’s most wicked problems. Perhaps with AI as our social change partner, we can see rays of light, in what is otherwise a heavy world for many, entrenched in scarcity, fear and lack.
Hope and excitement for AI as our co-partner is something that could be catalyzing. So, I am starting to think of AI as a much needed accelerator to help us solve complex world problems, in ways that could have taken us decades or would not have been possible at all. With AI co-partnering on local, regional and globally scaled solutions, lives and industries are evolving in new directions, with real impact and outcomes.
This way of being with technology is perhaps what can unlock our potential to see technology as ours to wield for the highest and best purposes. Responsible innovation cannot just be a slogan, but an active practice.
But when we talk about AI, what are we referring to? The choice of AI technology depends entirely on the nature of the social problem and the specific solution being developed.
- Large Language Models (LLMs): A type of artificial intelligence (AI) model designed to understand, process, and generate human-like text. They are trained on massive amounts of text data, allowing them to learn the nuances of language, grammar, and context. They enable chatbots and virtual assistants for content creation and language translation or even supporting parents through maternal health.
- Computer Vision: If the problem involves analyzing images or videos (e.g., detecting hate speech in online content, monitoring traffic patterns for urban planning), computer vision algorithms are used
- Reinforcement Learning: For problems that require an AI to learn through trial and error and make decisions in complex environments. Examples of where RL could be useful include, optimizing resource allocation in a social welfare program, controlling traffic lights for better flow or developing assistive technologies for those with disabilities
- Recommendation Systems: When the goal is to personalize information or services to individuals. Examples include matching job seekers with suitable opportunities, suggesting educational resources based on individual needs.
- Predictive Modeling: If the problem involves forecasting future trends or events (e.g., predicting crime hotspots, identifying individuals at risk of health issues), predictive modeling techniques would be employed.
What is AI changing?
To understand the unique and deeply beneficial ways AI is impacting our lives, we can’t just look at what AI does. We have to understand how it changes the trajectory of a specific problem, need or gap, and creates new outcomes. Some questions we can start asking about AI as our co-partner in solving wicked problems are:
- How can AI help us identify and create solutions for root level problems?
- What can AI see and do that we don’t have the current skills for?
- How can AI change how a field, industry, service, consumer experience, product evolve in new timeframes, efficiencies, enabling discoveries, solutions and data, not possible before?
Here are social challenges that AI is impacting, across the globe.
Health Outcomes
As health problems become more complex, it can be difficult to find solutions. What if AI could apply skills to enhance our potential for solutions, accelerate the time in which we find solutions, and access and engage with data and societal contexts, in ways that we currently cannot?
A recent study published in JAMA Oncology estimates that the total cost of cancer to the global economy will reach $25.2 trillion between 2020 and 2050. The stress of cancer on patients, communities, and healthcare systems is real. Using AI in healthcare is changing people’s health trajectory. Most recently, an AI powered diagnostic tool was able to diagnose a rare form of leukemia, saving the life of a 60 year old woman in Japan. She had been originally diagnosed and treated with another form of leukemia, without any positive results. The medical team used IBM’s Watson, an AI-powered data analytics system that can understand natural language, process massive amounts of data, and learn from it. Watson came to the diagnosis in 10 minutes. Without this diagnosis, the woman would not have made it.
In India alone, it is estimated that 29.8 million Indians will suffer from cancer. Swasthya.ai, a health tech startup founded by Akshay Navalakha and Rahul Jeshnani, recognizes the stress, loneliness and burden of going through such a health challenge. They have created a personalized AI enabled platform that brings insight from patient histories, oncology expertise, and other data to ensure patients receive hightouch interventions that address their particular needs and challenges. In a country with such a huge population, using AI could be a game changer in terms of addressing personalized care solutions that would otherwise be labor intensive.
Dr. Dionne Mahaffey is changing the experience of mental health practitioners and their patients. She’s created Behavior Health Notes, an AI powered platform that captures notes from therapy sessions, refocusing where practitioners spend their time and energy. It might not seem like a big deal, but therapists documenting their sessions can lead to burn out for various reasons, including enduring the emotional toll of reliving the trauma of their patients, the administrative burden and the challenge of documenting and taking accurate notes. Dr. Mahaffey created this platform based on her own experiences of documentation. Her intervention is significant in a healthcare field that is in high demand with the costs of mental health reaching over $1 trillion globally. A user of the platform said, “As someone who’s been in the field for decades, I’ve seen technology evolve, but never something this revolutionary for mental health clinicians. This AI system has turned the tedious task of note-writing into a breeze. I’m now saving at least 10 hours a week, and that’s not an exaggeration”
These solutions are changing health outcomes for patients and professionals. They are also changing how healthcare systems solve complex healthcare needs, increasing efficiency, and opening up possibilities to focus on human centered care.
Educational Outcomes
Educational inequity is a complex problem that spans educational policy, investment, skills gaps, lack of qualified teachers, lack of quality curriculum and pedagogy, and more. UNESCO estimates that 250 million children and young people around the globe are not in school and 70% of 10 year olds in low and middle income countries do not have basic literacy skills. How is AI being used to address these issues?
Founded in 2011 by Claudio Sassaki and Eduardo Bontempo, Geekie is a world bank recognized AI educational platform that creates personalized learning for children in Brazil. They are changing the quality of education by ensuring that each child is nurtured with the skills and knowledge that will set them up for the future. Over half of Brazilian students dropout of high school. We know that educational access and background play into economic and social mobility in the future. So Geekie is solving a complex problem that has generational impacts. Geekie is using AI to assess individual learning needs and develop learning plans that meet student needs. They have impacted 12 million students so far and are recognized as increasing educational outcomes scores and developmental gains by the World Bank. “Studies carried out by the consultancy METAS Sociais proved the impact of the platform: students from public schools who used the platform obtained, on average, an increase of 72 points in the official simulation of the MEC. Students who completely followed our personalized study plan had a five times greater development than those who did not benefit from this adaptive approach.” They believe AI has unlocked their ability to address a macro level problem, through a micro level solution using AI data analytics and ability to build solutions that would otherwise require high investment resources in underinvested schools and educational systems.
Literacy, however, is not just an educational access and quality issue. There is also the challenge of being technologically literate in a world that is linguistically uneven. About 1.5 billion people are fluent in English, which is about 18% of the world. Yet most data, tools and frameworks, and research and development happens in English, and in English speaking countries, excluding most of the world from accessing AI’s potential to address their vital social challenges. Vambo AI, created in 2019, and founded by Dzinotyiwei and Isheanesu Misi, aims to solve the access problem in the African continent. They are using AI to help people write emails, access information on the internet, and translate texts in a language of their choice. They are teaching people how to use AI in their language, resourcing them with the ability to generate content, and taking part in economic opportunities AI could make possible throughout the African continent.
Wealth Outcomes
Wealth inequity is growing globally. Among industrial nations, the United States is by far the most top-heavy, with much greater shares of national wealth and income going to the richest 1 percent than any other country, according to Inequality.org. The racial wealth gap in the United States has grown over the past four decades. The gap in home ownership for Black Americans is 30%, according to the Treasury Department, and Black applicants are 1.8x more likely to be denied mortgages, according to The Markup.
Stratyfy is working with financial institutions to ensure they are making unbiased financial decisions. They are using AI predictive analytics to work with lending financial institutions to ensure people of color are not denied loans due to systemic racism. They empower these institutions to assess lending risk and creditworthiness through alternative data, rather than traditional sources like zip codes and FICO scores. They also ensure algorithms based on such data are not deciding lending outcomes. They are currently working with 20 lenders and are aiming to impact 14.9 million households. Co-founders Laura Kornhauser and Dmitry Lesnik recognized the constant trade-offs they had to make between accuracy and transparency in the financial world, so they developed a vision for “AI-based decision making — one where AI-enabled data insights and human understanding were combined to embrace the complexity of how important decisions really needed to be made.” The opportunity here is to harness other ways of looking at who can access wealth generation and how.
Paycode is a fintech company using AI to ensure unbanked and underserved communities have access to financial tools and resources in the African continent. They are giving people in deep rural areas the ability to make financial transactions in real time and offline, solving a compelling need. Around the world, 1 billion have no formal identity, 3.7 billion don’t have any connectivity and 1.7 billion people remain unbanked, according to Paycode. Without access to financial services, people could be leaving money in places that are vulnerable to theft or unable to build a financial identity, impacting credit, access, mobility and building of financial assets. Small farmer Lisa Kabumbya, from Chikankata, Zambia, said: “Through these cards, I have improved my farming. I can sell my crops and get the money on my card immediately. My money is safe all the time. Even when I am in the bush far away, I am able to purchase goods using the card. It also helps me as it doesn’t require a password to transact, but I just use my fingerprints. Also, even though there is no internet, we can still do transactions.”
Food Supply and Access Outcomes
The global food production system is challenged by social, economic and environmental forces contributing to food waste, insecurity, hunger and malnutrition. The cost of a healthy diet is out of reach for 3 billion people worldwide. Addressing this global challenge requires a multi-faceted approach, including investments in agriculture, food systems, social safety nets, and public health interventions. How is AI helping?
Sprk.global is an AI driven platform that is reducing food waste, while also decreasing carbon emissions. They are using AI to identify patterns in the food supply ecosystem, including where there is food surplus. They state: “The first step to driving down food waste is to build inventory and visualize the available type and amount of surplus. We then add preferences and surplus information to the structured data and let algorithms optimize the matches. Always with one goal: facilitating the rapid distribution of surplus food of the supply chain while minimizing the CO2 footprint.”
A new project called Nourish is using AI to address the issue of food deserts across the United States. Food desserts are places where communities don’t have access to quality food, impacting health and economic outcomes. It’s estimated that 39 million people in the United States do not live within an accessible distance to quality food. The Nourish app will make it possible to pair small businesses and community based organizations to be connected to food growers, giving them resources to make fresh and quality food available to their communities. Leads on the project are excited about how AI can help bring the various elements of the food system together to change food access and integration. According to Amarnath Gupta, one of the leads, “the problem is complex and requires the assimilation of a wide variety of information from governmental, business and private sources. We are approaching it as a knowledge-based recommendation challenge, and we are developing a number of information integration, natural language processing, graph analytics and conversational AI techniques to offer a comprehensive solution for several user categories.”
Conclusion
Social challenges around the world are not just historical but are compounding. This is where AI could unlock solutions in ways that are not possible without technological interventions. In order for the potential of AI to be our co-partner across the globe, we must address the complications of using AI responsibly and ethically. One baseline reality is that large language models, AI frameworks, and technological resources are still being built and housed largely in the United States, and with data sources that leave out most of the world. There are several dynamic shifts we will need to see, and more collective global resourcing by:
- Reducing the cost of building and using AI as a tool, which can be prohibitive and expensive
- Ensuring data sources that train AI models represent more of the world
- Solutions and products created in the global south or by smaller startups have access to resources, funding and technologies that allow them to own their solutions and redistribute the outcomes in their communities and contexts.
- Eliminating the integration of negatively biased data and inequitable societal contexts into the entire lifecycle of AI development, use and adoption.
In a world where AI is our co-partner, can we envision a paradigm shift where social challenges transition to universal access, boundless opportunities, and a more humane society for all? Actively engaging with the inherent complexities and ethical realities surrounding AI, we are still at a new moment for global transformation. If we harness this potential responsibly and equitably, we may collectively shape a future where all of humanity thrives and flourishes.
Your moonshot maven, Dr. Salima Bhimani
Connect with me on LinkedIn
Listen to our podcast Global Code
