Clearing the way for participatory data stewardship in artificial intelligence development: a mixed methods approach

Area of Science:

• Computer Science
• Human-Computer Interaction
• Social Science

Background:

• Artificial intelligence (AI) development relies on vast datasets, necessitating ethical data sourcing.
• Participatory data stewardship (PDS) offers a framework for individuals to control their data in AI.
• Understanding consumer willingness to contribute data to AI is crucial for responsible AI.

Purpose of the Study:

• To identify key factors influencing consumers' willingness to provide data for AI via PDS.
• To extend the Technology Acceptance Model (TAM) with trust and subjective norms.
• To assess the impact of social duty, purpose understanding, and guilt on data donation.

Main Methods:

• A mixed-methods study utilizing an experimental survey.
• Recruited 322 participants to assess their data donation profiles.
• Applied an extended TAM incorporating trust, subjective norms, social duty, purpose understanding, and guilt.

Main Results:

• Trust in AI, understanding the AI's purpose, and a sense of social duty significantly predicted willingness to provide data.
• The TAM, even when extended, may not fully capture all factors influencing user willingness.
• Participants prioritized trust and comprehension of AI's societal impact when donating data.

Conclusions:

• Consumer willingness to engage in PDS for AI is driven by trust, perceived social duty, and clarity of purpose.
• Future research should explore additional factors beyond the TAM to fully understand data donation behavior.
• Fostering trust and transparency is essential for encouraging participation in AI data stewardship.