Institute for Manufacturing, Engineering Department, U. of Cambridge, United Kingdom
Decision-making is a fundamental yet highly costly task for humans. This complex process is inherently nested and iterative. Artificial Intelligence (AI), a rapidly evolving technology, is expected to increasingly support human decision-making. However, we have yet to fully understand how the adoption of this emergent technology will impact humanity. AI’s unique characteristics—its agentic and probabilistic nature, lack of physical embodiment, high connectivity, and non-observability—distinguish it from previous decision-support tools. In this context, we conceptualize three key problems that are crucial for future discussions on integrating AI into human decision-making: 1) The “Theory of Mind” Problem which sees an imbalance in the way AI and humans understand each other; 2) The “Control” Problem which relates to the extent to which humans can retain control over AI systems; and 3) The “Impact” Problem pertaining to an evaluation of the consequences of AI integration for individuals, organizations, and society. We further discuss the larger implications of these problems and propose directions for future research. Our aim is to ensure that humans fully benefit from AI’s capabilities while mitigating potential negative impacts.
