AI-Augmented Prevention Science Needs Community-Engaged Prevention Science: a Framework for Greater Accountability

Area of Science:

• Prevention Science
• Artificial Intelligence
• Community Engagement

Background:

• Artificial intelligence (AI) offers potential for addressing societal challenges but can yield biased and discriminatory results.
• Existing AI research frameworks often lack accountability to the communities impacted by these technologies.

Purpose of the Study:

• To present a model of community-engaged AI-augmented prevention science.
• To integrate community input throughout the prevention science lifecycle to mitigate AI-related harms.
• To address racial biases in AI algorithms used for social interventions.

Main Methods:

• Developed a prevention science framework for community-engaged AI research.
• Highlighted roles for community and AI at each stage of the prevention science lifecycle.
• Integrated AI and community-engaged methods to correct racial biases in a predictive risk algorithm for housing interventions.

Main Results:

• The proposed model emphasizes community input for critical decision-making moments in AI development.
• Case study demonstrated successful correction of racial biases in a predictive risk algorithm.
• Community-informed "critical retreats" led to more equitable and inclusive intervention outcomes.

Conclusions:

• Community-engaged approaches are essential for developing responsible and equitable AI in prevention science.
• Future research should focus on advancing methods for integrating community values into AI.
• AI-augmented prevention science must prioritize accountability to the communities it serves.