Intelligent Systems For Good

The United Nations’ 2030 Agenda for Sustainable Development defines 17 Sustainable Development Goals (SDGs) that address humanity’s most pressing challenges: from climate action and responsible consumption to decent work, and cultural preservation. While technological innovation is often discussed in the context of economic growth, its potential to meaningfully contribute to these broader sustainability goals remains underexplored. This thesis argues that Intelligent Systems must go beyond traditional metrics of accuracy and computational performance to explicitly consider how technology can advance societal well-being across environmental, social, economic, and cultural dimensions. This thesis demonstrates how a human-centered approach to Intelligent Systems design and evaluation can address four specific SDGs (8, 11, 12, 13) through complementary research threads, each tackling a different dimension of technology for social good. The unifying theme across all threads is that contribution to the SDGs requires not only technical innovation but also rigorous evaluation that combines offline metrics with online user studies to ensure that systems deliver actual benefits as experienced by real users. Addressing SDG 12 (Responsible Consumption and Production) and SDG 13 (Climate Action), we investigated two complementary approaches to environmental sustainability, both aiming to influence user behavior and to reduce the computational footprint of recommendation. We tested our approaches with two user studies. Addressing SDG 8 (Decent Work and Economic Growth), we investigate how AI autonomy affects productivity, trust, and the quality of work, through a user study. For SDG 11 (Sustainable Cities and Communities), particularly the goal to safeguard cultural heritage, we developed two complementary systems to make cultural heritage information more accessible, and validated our findings with user studies. Finally, recognizing that accountability is fundamental to achieving all SDGs, we investigated reproducibility in a human-centered approach. To conclude, Intelligent Systems can meaningfully contribute to the Sustainable Development Goals. However, doing so requires expanding our definition of success beyond accuracy and efficiency to encompass environmental sustainability, economic prosperity, cultural preservation, and accountability. Through human-centered evaluation combining offline metrics with online user studies, we can ensure that technological progress genuinely serves the greater good rather than merely claiming to do so.