Fair contracts for artificial intelligence?

Area of Science:

• Artificial Intelligence
• Philosophy of AI
• Ethics in AI

Background:

• The alignment problem in artificial intelligence (AI) seeks to ensure AI systems act in accordance with human values.
• Philosophical contractualism offers frameworks for ethical decision-making based on agreements.
• Resource-rational contractualism (RRC) is an economic theory of bounded rationality.

Purpose of the Study:

• To explore resource-rational contractualism (RRC) as a viable framework for AI alignment and reasoning.
• To assess the applicability of RRC to AI by evaluating its alignment with philosophical contractualism.
• To propose a modified RRC approach for guiding AI moral cognition and alignment.

Main Methods:

• Conceptual analysis of resource-rational contractualism (RRC).
• Assessment of RRC's alignment with core tenets of philosophical contractualism.
• Extension of RRC principles to the domain of artificial intelligence alignment.

Main Results:

• Resource-rational contractualism (RRC) demonstrates potential as a framework for AI alignment.
• A modified RRC, emphasizing impartial deliberation, can effectively capture key insights from philosophical contractualism.
• This modified RRC can identify alignment principles and direct efficient AI moral cognition.

Conclusions:

• Resource-rational contractualism offers a promising, albeit modified, approach to artificial intelligence alignment.
• Impartial deliberation within RRC is crucial for developing effective AI ethical reasoning.
• The framework provides a pathway for creating more aligned and morally cognisant AI systems.