Harnessing AI for a Better World
There are many examples of using AI in the wrong way, as highlighted in the thought-provoking book Weapons of Math Destruction. Also, the risks of AI should not be underestimated. AI ethics and governance have become pressing concerns in our rapidly evolving technological landscape, and numerous companies now maintain dedicated departments focused solely on addressing these issues. However, amid these challenges, there are also inspiring stories that demonstrate how mathematics and AI can be harnessed for the greater good.
This post is written to inspire you and to show how AI can help addressing complex global challenges, from climate change to human rights abuse. The selection of initiatives and companies managed to achieve remarkable results on a large scale and I do believe they deserve a spotlight. It’s only a selection, because there are many more examples available. If you come across something I’ve missed, please don’t hesitate to comment on this post!
One additional motivation for sharing this post is World Animal Day, celebrated on October 4th. 🐈 ⬛
Refugee Resettlement
Mathematical optimization is typically applied to maximize profits or minimize costs in various scenarios. However, there are exceptional cases where it serves a noble purpose, as demonstrated here.
When refugees seek shelter in a host country, the initial placement of a refugee family into a suitable home can profoundly affect their long-term prospects in terms of employment, education, and overall well-being. Many refugees are vulnerable and it’s important to place them in a safe environment where they can start building a new live. Similarly, for foster homes, it’s essential to align their preferences with the refugees they welcome. For instance, if an elderly woman desires to care for refugees but ends up hosting a family of five men, the situation can be overwhelming. By optimizing the matchmaking process, it’s possible to significantly improve placement outcomes.
Annie MOORE is a software system developed in 2018 that focuses on this problem. The software is used by HIAS, a resettlement agency based in the United States. Annie uses machine learning and mathematical optimization to recommend matches between foster homes and refugees. This software contributed to an impressive increase in employment outcomes for refugees resettled by HIAS, with figures ranging from 22% to 38%.
Taking this innovative approach a step further, a new platform called RUTH (Refugees Uniting Through HIAS) enhances the resettlement process. RUTH incorporates the preferences of both refugees and host families, making relocation faster and more transparent. Dr. Trapp, Associate Professor of Operations and Industrial Engineering at Worcester Polytechnic Institute, highlights the significance of RUTH by stating:
"This is the first time ever that preferences of refugees and priorities of hosts have been systematically used in the resettlement process."
RUTH helped to place refugees from Ukraine in the US during the Russian invasion.
Safeguarding Wildlife
AI is playing a vital role in the protection and conservation of endangered species and wildlife. Leveraging advanced image recognition and predictive modeling, AI-powered systems have revolutionized the way we monitor animal behavior, track migration patterns, and combat poaching threats (yes, poachers still exist). These technologies empower conservationists and law enforcement agencies to respond promptly, safeguarding the world’s biodiversity and preserving our natural heritage.
There are many remarkable examples that can be found on the web. Let’s go through some of them.
Saving species from extinction is something that computer vision and other machine learning techniques can help with. A possibility to keep track of species is to count animals in national parks. Appsilon developed a tool called Mbaza AI. This tool is basically an image classification tool specialized in wildlife. An endangered species in Australia is the koala, because of bushfires and animal attacks. This initiative uses AI to locate and rescue surviving koalas. Collaborating with the Snow Leopard Trust, Microsoft AI is engaged in the detection and recognition of snow leopards, aiding in their conservation and protection. Google AI uses the songs of whales to locate and protect them.
Unfortunately, worldwide there are still poachers hunting down animals. But luckily, AI is making their job way more difficult. Hack the Planet has an initiative that can alert rangers if a poacher is on the road. This is great, because the ranger can go after the poacher directly, instead of watching camera footage constantly. Another project monitors every boat that goes in and out of a park in Zambia, to prevent illegal fishing. Other animals that are monitored with the help of AI are grizzly bears, elephants and penguins.
Combatting Climate Change
Climate change is a big issue. We have to deal with the negative effects of climate change, like rising sea levels, extreme weather events, heightened health risks, food scarcity and population displacement.
The battle against climate change requires data-driven insights and sustainable solutions. AI excels in processing vast datasets and identifying patterns that inform climate policies and resource management. From optimizing renewable energy production to predicting extreme weather events, AI contributes to mitigating the effects of climate change and promoting a more sustainable future for our planet.
Accurate weather forecasting is vital for protecting lives and property. LEAP (Learning the Earth with AI and Physics) is a science technology center dedicated to improve short-term climate forecasts. They achieve this by blending traditional climate science with advanced machine learning. LEAP combines the strengths of both methods, and builds next generation AI models.
As you might know, carbon emissions have an impact on the planet. Many initiatives show that it’s possible to reduce them. A company that uses AI to reduce the carbon footprint is Eugenie.ai, that helps manufacturers to decarbonize their operations. Mortar IO discovers ways to decarbonize existing buildings, their vision is to be the data infrastructure for decarbonizing real estate. The carbon footprint of buildings is equal to 39% of global energy related carbon emissions, so this is quite a good initiative.
An interesting question that might come to mind: What about the emission footprint of AI itself? AI is quite a big consumer (visual in the first chart of this post), and it is possible to reduce the emission drastically, as HuggingFace proved with the BLOOM model. New technologies can help reducing the carbon footprint, also from AI. Maybe the best way to lower the emission is to work on smaller models that perform as good as the bigger ones.
Waste, another topic that arises when you think of climate change. The consumption behavior of people has a huge impact on methane emissions. A cool initiative that handles waste from the fashion industry is Refiberd. They use AI to sort textiles for recycling applications.
Helping Farmers in Developing Nations
In developing nations, small-scale farmers often face challenges in maximizing crop yields and managing resources efficiently. AI-driven applications provide valuable support by offering tailored advice on crop selection, irrigation, and pest control. By equipping farmers with these tools, we empower them to improve their livelihoods and contribute to food security in their communities.
An application that is used al over the world to detect diseases of crops and provide solutions is Plantix. It helps maximizing crop yields and uses AI to detect them. Hello Tractor is a company that provides a platform connecting farmers with tractor owners for plowing and other farming activities. They use mathematical optimization and machine learning to optimize tractor allocation and scheduling, making it more affordable for smallholder farmers in developing nations to access mechanized farming. Another concrete example is Apollo Agriculture. This company helps farmers in Kenya and Zambia, not only with farming, but also with financing. The machine learning part of Apollo focuses on credit models to make credit decisions.
There are many more initiatives, like AgriPredict, Taranis and Farmshine (platforms for farmers with weather predictions, crop disease diagnoses and market access).
Exposing Human Rights Abuses and War Crimes
Many human rights organizations fight AI because of the risks of it. Check out recent posts from Amnesty International, Human Rights Watch and Bellingcat. Besides the criticism, can AI play a positive role in this field as well?
Actually, it can: AI can sift through vast amounts of data, including images, videos, and textual information, to identify and document violations. This is done through image and video analysis, NLP, facial recognition and predictive analytics. Also, AI can aggregate and cross-reference data from various sources to build a comprehensive picture of events and human rights violations. This includes collating information from eyewitness accounts, social media posts, and official reports.
Syrian Archive is an organization that uses open-source intelligence, digital forensics, and AI to document human rights violations in Syria. They collect and verify visual evidence, such as photos and videos, to hold perpetrators accountable. Here you can read more on the methods and tools they use.
Another example is Forensic Architecture. This research agency uses spatial and architectural analysis, combined with AI and machine learning, to investigate human rights violations and state violence. They often collaborate with other organizations to provide valuable insights. These examples show different use cases in which machine learning techniques were utilized.
And last but not least, the Centre for Information Resilience is a non-profit social enterprise that is dedicated to exposing human rights abuses and war crimes. Besides, they also counter disinformation and combat harmful online behavior. An example in which they use data science (mostly analysis and visualization) is the Eyes on Russia map. It is a timeline which outlines important events since the Russian invasion of Ukraine.
Hopefully more initiatives like these will arise to use Ai For Good.
Conclusion
While the ethical challenges of AI are undeniable, it is essential to recognize its potential for creating positive change in the world. The examples discussed here demonstrate that, when harnessed responsibly and with a commitment to ethical principles, AI and mathematics can be powerful tools for addressing pressing global issues and making the world a better place.
If you like to volunteer or work for one of the projects and companies, some of them do have career opportunities! You can go to their website and search for it.
Related
Ethical Considerations In Machine Learning Projects
Five ways to combine Mathematical Optimization and Machine Learning
Model-Agnostic Methods for Interpreting any Machine Learning Model
