From Exploitation to Augmentation: Navigating AI’s Impact on Human Value- and Inference-based Decision-Making
conference paper
Multi-Domain Operations (MDO) drastically increase the complexity of military decision-making. Artificial Intelligence (AI) has the potential to augment decision-making processes and assist human decision-makers in dealing with this growing complexity. However, AI also poses a risk, as influential algorithms can be exploited by adversaries to influence perceptions, beliefs and behaviours and to manipulate public opinion to gain strategic advantages (e.g., in cognitive warfare). To understand AI’s effects on human decision-making, we distinguish between two types of decision tasks: inference-based decisions (focused on situational understanding) and value-based decisions (focused on choosing a course of action). In the present paper, we argue that AI has differential effects on both types of decisions, augmenting and exploiting them. AI can enhance inference-based decisions by its computational power, but it can also manipulate human judgments by facilitating the spread of misinformation. For value-based decisions, AI can help elicit preferences and calculate optimal options to achieve military goals. However, AI-driven profile-based, micro-targeting techniques can be used to manipulate people into choices that are inconsistent with their true values. We aim to offer insights into how to best navigate these influences to mitigate the risks and to maintain meaningful human control.
TNO Identifier
1014278
Source title
NATO HFM-RSY-377 Symposium on Meaningful Human Control of Future Military Operations 2024: Spanning Across Warfare Domains with Advanced AI - 21 Oct-22 Oct, Amsterdam, Netherlands