Artificial Intelligence for Social Good: A Survey (2001.01818v1)

Abstract: Artificial intelligence for social good (AI4SG) is a research theme that aims to use and advance artificial intelligence to address societal issues and improve the well-being of the world. AI4SG has received lots of attention from the research community in the past decade with several successful applications. Building on the most comprehensive collection of the AI4SG literature to date with over 1000 contributed papers, we provide a detailed account and analysis of the work under the theme in the following ways. (1) We quantitatively analyze the distribution and trend of the AI4SG literature in terms of application domains and AI techniques used. (2) We propose three conceptual methods to systematically group the existing literature and analyze the eight AI4SG application domains in a unified framework. (3) We distill five research topics that represent the common challenges in AI4SG across various application domains. (4) We discuss five issues that, we hope, can shed light on the future development of the AI4SG research.

• The paper presents a comprehensive review of AI applications across diverse social domains by aggregating insights from over 1,000 studies.
• It categorizes the literature using three conceptual frameworks and highlights healthcare as the dominant focus with 32% coverage in 2019.
• The survey identifies common challenges such as limited data, privacy concerns, and the need for human-centered, sustainable AI deployment.

Overview of "Artificial Intelligence for Social Good: A Survey"

The paper "Artificial Intelligence for Social Good: A Survey" provides a comprehensive review of the burgeoning domain of Artificial Intelligence for Social Good (AI4SG). Authored by Zheyuan Ryan Shi, Claire Wang, and Fei Fang from Carnegie Mellon University, it aggregates insights from over 1000 papers to elucidate the strides made in applying AI to ameliorate societal issues. This survey meticulously categorizes and analyzes the literature by application domains and AI techniques, proposing a structured framework to further explore this domain. It also identifies critical research challenges that are common across various domains and outlines potential future directions for AI4SG.

Literature Distribution and Trends

The survey reveals a significant increase in AI4SG research over the past decade, with healthcare receiving the most attention, accounting for 32% of the literature in 2019. Transportation and environmental sustainability also emerge as prominent fields within AI4SG, whereas agriculture, education, and social care receive comparatively less focus. This disparity is attributed to factors such as the maturity of research fields and the availability of structured data. Furthermore, Machine Learning techniques dominate the AI4SG landscape, often integrated into other AI methodologies like computer vision and natural language processing.

Conceptual Frameworks and Domain-Specific Applications

The authors introduce three conceptual frameworks to systematically categorize AI4SG problems: the AEC model (Agent – Environment – Community), the DPP model (Descriptive – Predictive – Prescriptive), and the domain-specific topic structure. These frameworks help elucidate potential AI interventions across eight identified application domains: agriculture, education, environmental sustainability, healthcare, information manipulation, social care, public safety, and transportation.

Case Studies and Real-World Deployment

The paper includes several case studies that exemplify successful AI deployments, such as optimizing patrol routes to combat wildlife poaching and improving water pipe replacements in Flint, Michigan. These examples highlight the practical utility of AI in these domains and identify key factors contributing to successful deployment, such as collaboration with domain experts and stakeholders, data availability, and overcoming logistical challenges.

Research Progress and Challenges

The survey identifies several common research challenges inherent to AI4SG, including limited data availability, data bias, privacy concerns, human-computer interaction, and sustainability of AI deployments. These challenges necessitate advanced methodological developments, such as transfer learning, differential privacy, and robust deployment strategies. Moreover, the paper emphasizes the importance of human-centered AI approaches to ensure that AI systems are valuable and ethical components of decision-making processes.

Implications and Future Directions

The findings in the survey underscore the transformative potential of AI4SG initiatives in addressing global problems while also highlighting the necessity for rigorous evaluation and sustainable implementation of AI systems. The authors argue for increased collaboration between academia and industry to leverage resources and expertise, as well as the need for more inclusive research efforts that promote diversity in AI development.

This survey serves as a cornerstone for researchers exploring AI4SG by providing a structured understanding of current trends, challenges, and opportunities in this impactful field. It calls for a concerted effort to harness AI’s capabilities for societal benefit while ensuring ethical and equitable advancements.