AI for Good: Applications in Sustainability, Humanitarian Action, and Health

FOREWORD BY BRAD SMITH, VICE CHAIR AND PRESIDENT OF MICROSOFT Discover how AI leaders and researchers are using AI to transform the world for the better In AI for Good: Applications in Sustainability, Humanitarian Action, and Health , a team of veteran Microsoft AI researchers delivers an insightful and fascinating discussion of how one of the world's most recognizable software companies is tackling intractable social problems with the power of artificial intelligence (AI). In the book, you’ll see real in-the-field examples of researchers using AI with replicable methods and reusable AI code to inspire your own uses. The authors also provide: Easy-to-follow, non-technical explanations of what AI is and how it works Examples of the use of AI for scientists working on mitigating climate change, showing how AI can better analyze data without human bias, remedy pattern recognition deficits, and make use of satellite and other data on a scale never seen before so policy makers can make informed decisions Real applications of AI in humanitarian action, whether in speeding disaster relief with more accurate data for first responders or in helping address populations that have experienced adversity with examples of how analytics is being used to promote inclusivity A deep focus on AI in healthcare where it is improving provider productivity and patient experience, reducing per-capita healthcare costs, and increasing care access, equity, and outcomes Discussions of the future of AI in the realm of social benefit organizations and efforts Beyond the work of the authors, contributors, and researchers highlighted in the book, AI For Good begins with a foreword from Microsoft Vice Chair and President Brad Smith. There, Smith details the Microsoft rationale behind the creation of and continued investment in the AI for Good Lab. The vision is one of hope with AI saving lives in disasters, improving health care globally, and Microsoft's mission to make sure AI's benefits are available to all. An essential guide to impactful social change with artificial intelligence, AI for Good is a must-read resource for technical and non-technical professionals interested in AI’s social potential, as well as policymakers, regulators, NGO professionals, and non-profit volunteers. Cover Title Page Copyright Page Contents Foreword Introduction A Call to Action Part I Primer on Artificial Intelligence and Machine Learning Chapter 1 What Is Artificial Intelligence and How Can It Be Used for Good? What Is Artificial Intelligence? What If Artificial Intelligence Were Used to Improve Societal Good? Chapter 2 Artificial Intelligence: Its Application and Limitations Why Now? The Challenges and Lessons Learned from Using Artificial Intelligence Models Can Be Fooled by Bias Predictive Power Does Not Imply Causation AI Algorithms Can Discriminate Models Can Cheat (the Problem with Shortcut Learning) Models Do Not Generalize to Out-of-Distribution Cases Models Can Be Gamed Some Tools Can Be Used as Weapons Models Can Create an Illusion of Certainty AI Expertise Alone Cannot Solve World Problems Conclusion Large Language Models Understanding Language Models The Training Process: Learning Language Through Data Historical Perspective: Two Decades of Evolution The Generative Aspect of GPT Pre-training: The P in GPT and Beyond Transformers: The T in GPT and Its Revolutionary Impact Limitations of LLMs Demystifying AI’s Intelligence Understanding Truth The Phenomenon of LLM Hallucinations The Impact of LLMs LLMs and the Power for Good LLMs as a Language Aid LLMs for Democratizing Coding LLMs in Areas Like Medicine Chapter 3 Commonly Used Processes and Terms Common Processes Commonly Used Measures The Structure of the Book Part II Sustainability Chapter 4 Deep Learning with Geospatial Data Executive Summary Why Is This Important? Methods Used Findings Discussion What We Learned Chapter 5 Nature-Dependent Tourism Executive Summary Why Is This Important? Methods Used Findings Discussion What We Learned Chapter 6 Wildlife Bioacoustics Detection Executive Summary Why Is This Important? Methods Used Findings Discussion What We Learned Chapter 7 Using Satellites to Monitor Whales from Space Executive Summary Why Is This Important? Methods Used Findings Discussion What We Learned Chapter 8 Social Networks of Giraffes Executive Summary Why Is This Important? Methods Used Findings Discussion What We Learned Chapter 9 Data-driven Approaches to Wildlife Conflict Mitigation in the Maasai Mara Executive Summary Why Is This Important? Methods Used Findings Discussion What We Learned Chapter 10 Mapping Industrial Poultry Operations at Scale Executive Summary Why Is This Important? Methods Used Findings Discussion What We Learned Chapter 11 Identifying Solar Energy Locations in India Executive Summary Why Is This Important? Methods Used Findings Discussion What We Learned Chapter 12 Mapping Glacial Lakes Executive Summary Why Is This Important? Methods Used Findings Discussion What We Learned Chapter 13 Forecasting and Explaining Degradation of Solar Panels with AI Executive Summary Why Is This Important? Methods Used Findings Discussion What We Learned Part III Humanitarian Action Chapter 14 Post-Disaster Building Damage Assessment Executive Summary Why Is This Important? Methods Used Findings Discussion What We Learned Chapter 15 Dwelling Type Classification Executive Summary Why Is This Important? Methods Used Findings Discussion What We Learned Chapter 16 Damage Assessment Following the 2023 Earthquake in Turkey Executive Summary Why Is This Important? Methods Used Findings Discussion What We Learned Chapter 17 Food Security Analysis Executive Summary Why Is This Important? Methods Used Findings Discussion What We Learned Chapter 18 BankNote-Net: Open Dataset for Assistive Universal Currency Recognition Executive Summary Why Is This Important? Methods Used Findings Discussion What We Learned Chapter 19 Broadband Connectivity Executive Summary Why Is This Important? Methods Used Findings Discussion What We Learned Chapter 20 Monitoring the Syrian War with Natural Language Processing Executive Summary Why Is This Important? Methods Used Findings Discussion What We Learned Chapter 21 The Proliferation of Misinformation Online Executive Summary Why Is This Important? Methods Used Findings Discussion What We Learned Chapter 22 Unlocking the Potential of AI with Open Data Executive Summary Why Is This Important? Methods Used Findings Discussion What We Learned Part IV Health Chapter 23 Detecting Middle Ear Disease Executive Summary Why Is This Important? Methods Used Findings Discussion What We Learned Chapter 24 Detecting Leprosy in Vulnerable Populations Executive Summary Why Is This Important? Methods Used Findings Discussion What We Learned Chapter 25 Automated Segmentation of Prostate Cancer Metastases Executive Summary Why Is This Important? Methods Used Findings Discussion What We Learned Chapter 26 Screening Premature Infants for Retinopathy of Prematurity in Low-Resource Settings Executive Summary Why Is This Important? Methods Used Retinal Image Selector ROP Classifier and Model Calibration Mobile ROP Application Development Findings Discussion What We Learned Chapter 27 Long-Term Effects of COVID-19 Executive Summary Why Is This Important? Methods Used Findings Discussion What We Learned Chapter 28 Using Artificial Intelligence to Inform Pancreatic Cyst Management Executive Summary Why Is This Important? Methods Used Findings Discussion What We Learned Chapter 29 NLP-Supported Chatbot for Cigarette Smoking Cessation Executive Summary Why Is This Important? Methods Used Findings Final Version of QuitBot Quit Efficacy Randomized Controlled Trial Discussion What We Learned Chapter 30 Mapping Population Movement Using Satellite Imagery Executive Summary Why Is This Important? Methods Used Geographic Focus Building Density Estimated from Remote Sensing Data Estimating People per Structure Findings Discussion What We Learned Chapter 31 The Promise of AI and Generative Pre-Trained Transformer Models in Medicine What Are GPT Models and What Do They Do? GPT Models in Medicine Radiology Patient Self-Care Management and Informed Decision-Making Public Health Conclusion Part V Summary, Looking Forward, and Additional Resources Epilogue: Getting Good at AI for Good Communication Setting Realistic Expectations for AI Confronting Technical Limitations Project Scoping and Implementation Data Adapting to Previously Collected Datasets Creating Training and Test Sets with the Application Scenario in Mind Modeling Incorporating Domain Expertise Model Development with Resource Constraints Evaluation and Metrics Humans in the Loop Impact Uphill Path to Deployment and Adoption Measuring Impact Conclusion Key Takeaways AI and Satellites: Critical Tools to Help Us with Planetary Emergencies Amazing Things in the Amazon Quick Help Saving Lives in Disaster Response Additional Resources Endnotes Acknowledgments About the Editors About the Authors Microsoft’s AI for Good Lab Collaborators Index EULA