The AI Consumption Spiral: An Underestimated Systemic Risk for Economy and Society

Artificial intelligence is transforming economies at a depth that surpasses previous waves of automation. This paper examines whether its application generates a self-reinforcing AI consumption spiral. At the core lies the hypothesis that declining incomes weaken purchasing power, reduce demand, and lead companies to respond to this demand weakness with further automation. This creates a downward spiral that may threaten economic stability and social cohesion. Methodologically, the study combines theoretical model-building with a comprehensive literature review and draws on official data from international organizations. The findings indicate falling real wages despite rising nominal wages, increasing income inequality, and diverging consumption patterns. Particularly problematic is the loss of purchasing power among low-income households with high consumption shares, while wealthier groups with lower consumption propensity benefit. As a result, the demand base of the economy erodes. The socio-political analysis highlights that this shift reinforces inequality, intensifies polarization, and may impair the capacity of democratic institutions to act. Recent theoretical approaches that emphasize demand declines and expectations confirm the risk of a self-sustaining AI consumption spiral. Classical instruments such as retraining or basic income are insufficient to break the cycle. The paper concludes with the thesis that new politico-economic strategies are required to safeguard purchasing power, stabilize expectations, and thereby contain the risk of a consumption spiral triggered by AI.