Expectation Inflation of Artificial Intelligence: When Human Demands on AI Exceed Moral Design Capacity

What happens when the ethical and interpretive standards that organisations and societies have for artificial intelligence (AI) exceed what it was originally designed to do? This paper presents Expectation Inflation as a conceptual framework for explaining how ethical risk in AI arises not from technical failures but from inflated human expectations and the excessive transfer of moral responsibility. Expectation Inflation captures the widening gap between human expectations and the actual capabilities of AI systems, which lack intention or conscience. Drawing on perspectives from Science and Technology Studies (STS), behavioural economics, and moral philosophy, the paper describes this phenomenon using the Expectation Inflation Curve, which includes three stages: Rational Delegation, Normative Drift, and Moral Substitution. It introduces Moral Design Capacity (MDC) as the point at which the delegated authority ceases to be ethically valid. To realign expectations, the paper suggests Expectation Governance, a framework encompassing three elements: Delegation Boundaries (limits on moral authority), Expectation Auditing (tracking for exaggerated expectations), and Moral Accountability Indexing (re-establishing ethical ownership). These approaches aim to manage moral over-delegation effectively. The study reinterprets AI ethics as a matter of maintaining moral balance instead of solely correcting algorithms, and highlights the importance of leadership in sustaining this equilibrium. By redefining AI governance as the management of moral expectations, the paper establishes a new research direction for exploring the dynamics of expectations in ethical scenarios. This framework allows organisations to identify and address moral risks linked with AI implementation, foster ethical resilience, and preserve leadership integrity in data-driven decision-making.