What does AI4SocialGood mean and how can technology be used for societal and social benefit? Our author explores this question and presents some examples of such projects.

Technology in itself is neither good nor bad, and can be used for different purposes. The discourse on the possible risks and consequences of using technology, especially artificial intelligence, is therefore a central and important topic of research. At the same time, however, the research field AI4SocialGood, sometimes also called AI for Social Impact [1], has also established itself, which aims to make a positive contribution to social problems by means of technology. While technologies for searching in texts or automatic image recognition can certainly make a social contribution depending on the context, these general technologies are not meant by AI4SocialGood. The thematic field rather comprises targeted applications that address a specific social problem.

Augmented Intelligence and the UN Sustainable Development Goals

More and more applications of augmented intelligence are pursuing the goal of achieving added value for society. These projects are often based on the UN’s 17 Sustainable Development Goals (SDGs) [2] (see Figure 1), and address challenges in social or environmental science. For example, it has been shown that each of the SDGs is already addressed by at least one AI-based project [3]. In particular, there are many projects in the area of good health and well-being, while other areas such as affordable and clean energy, gender equality and no poverty have a handful of projects.

Figure: The UN’s 17 Sustainable Development Goals.

Interdisciplinary cooperation

At AI4SocialGood, computer scientists come together with experts from other subject areas to work on a common challenge. This can be challenging due to the different professional working styles and cultures and has led to the development of specific guidelines. A Nature study [4] has identified the most important aspects and hurdles for successful collaboration in the field of AI4SocialGood with the involvement of various experts: 1.For the field of artificial intelligence in general:

• Expectations: what is possible with AI?
• Simple solutions
• AI applications should be inclusive and accessible and comply with ethical principles and human rights, which must be regularly monitored

2. For the applications:

• Goals and use cases must be clear and well defined
• Deep and long-lasting partnerships are needed to successfully tackle large-scale problems
• Incentives must be aligned and the boundaries of the two communities must be taken into account in the planning process
• Building and maintaining trust is important to overcome organisational hurdles
• Opportunities to reduce the development costs of AI solutions should be thought through

3. And for data processing:

• Improving data provision is of great importance
• Data must be processed securely, with utmost respect for privacy and human rights

Example projects

Numerous projects already exist in the area of AI4SocialGood. The following list is not evaluative or conclusive, and is intended to give an impression of the contribution that augmented intelligence is already able to make in the social sphere:

• Amnesty International’s Troll Patrol project [5] used crowdsourcing, data science and machine learning to evaluate how much and in what form women are sexually harassed on online platforms.
• Machine learning has also been used to make predictions about poverty using satellite imagery [6]. This is helpful because reliable information about the economic situation in developing countries is often lacking but necessary to identify and combat poverty.
• In Microsoft’s Ocean Cleanup project, machine learning is used to automatically detect pollution such as plastic and simulate how it moves on the ocean. This allows the cleanup to be carried out more efficiently.

References

1. https://towardsdatascience.com/introduction-to-ai-for-social-good-875a8260c60f
2. https://www.un.org/development/desa/disabilities/envision2030.html
3. Cowls, J., Tsamados, A., Taddeo, M., & Floridi, L. (2021). A definition, benchmark and database of AI for social good initiatives. Nature Machine Intelligence, 3(2), 111-115.
4. Tomašev, N., Cornebise, J., Hutter, F., Mohamed, S., Picciariello, A., Connelly, B., … & Clopath, C. (2020). AI for social good: unlocking the opportunity for positive impact. Nature Communications, 11(1), 1-6.
5. https://decoders.amnesty.org/projects/troll-patrol/findings
6. Jean, N., Burke, M., Xie, M., Davis, W. M., Lobell, D. B., & Ermon, S. (2016). Combining satellite imagery and machine learning to predict poverty. Science, 353(6301), 790-794.