The case for near-term artificial intelligence risks to be considered a public health problem

Area of Science:

• Public Health
• Health Informatics
• Technology Ethics

Background:

• Artificial intelligence (AI) integration in healthcare is rapidly expanding.
• The dominant narrative focuses on AI's benefits, overlooking significant health risks.
• Near-term AI risks to health are emerging from current and near-future deployments.

Purpose of the Study:

• To outline prominent near-term AI risks to human health.
• To frame these risks as a public health problem.
• To propose risk management strategies for AI in healthcare.

Main Methods:

• Identification and categorization of near-term AI risks to health.
• Analysis of potential population-level health impacts.
• Review of existing discourse and counterarguments.

Main Results:

• Key risks include algorithmic bias, diagnostic errors, health disinformation, mental health impacts, unemployment, autonomous weapons, bioweapons, and cyberattacks.
• These risks are causing or could soon cause substantial population harm, meeting public health problem criteria.
• A benefit-focused narrative obscures these critical public health concerns.

Conclusions:

• Recognizing AI's near-term health risks as a public health issue is essential.
• Potential management strategies include mandatory AI health impact assessments and public health surveillance.
• Integrating public health expertise into AI governance is vital for mitigation.