AI-assisted rational decision-making

Area of Science:

• Decision Science
• Artificial Intelligence Ethics
• Cognitive Science

Background:

• Artificial intelligence (AI) is increasingly used to assist human decision-making across various domains.
• Understanding the extent and limitations of AI's utility in rational choice is crucial.
• Existing research often focuses on AI's information-gathering capabilities, but less on its role in rational justification.

Purpose of the Study:

• To analyze the ways and extent to which AI is useful for rational decision-making.
• To differentiate AI's utility across three distinct choice types: easy, hard, and transformative choices.

Main Methods:

• Conceptual analysis of AI's role in decision-making frameworks.
• Categorization of choices into easy, hard, and transformative based on epistemic and metaphysical properties.
• Evaluation of AI's contribution to information discovery, reason generation, and value maximization in each choice category.

Main Results:

• For easy choices, AI enhances efficiency and accuracy, increasing long-term value.
• For hard choices, AI assists in generating will-based reasons to commit to an option.
• For transformative choices, AI cannot bridge the epistemic or metaphysical gaps, thus not enabling rational decision-making.

Conclusions:

• AI's utility in rational decision-making is contingent on the nature of the choice.
• AI can augment, but not fundamentally enable, rational choice when the rational option is not discernible.
• The capacity for rational decision-making relies on the inherent values of options, which AI cannot create.