Multi-criteria consensus-based group decision making with prospect-regret TOPSIS for human-AI tool selection under...

Area of Science:

• Artificial Intelligence
• Decision Sciences
• Fuzzy Systems

Background:

• Rapid AI advancements necessitate frameworks for human-AI collaboration.
• Existing analytical frameworks struggle with uncertainty and imprecision in human-AI interactions.
• Linguistic Z-number (LZN) evolution models offer a promising approach to manage uncertainty and fuzziness.

Purpose of the Study:

• To extend multicriteria group decision-making (MCGDM) by incorporating linguistic Z-numbers (LZNs).
• To address challenges in MCGDM under LZNs, including consensus-building, expert weighting, and option ranking.
• To propose a practical, interactive consensus-building process allowing experts to revise opinions using original linguistic terms (LTs).

Main Methods:

• Developed an interactive, strategy-based consensus-building model for experts to revise opinions within original linguistic terms.
• Created an integrated subjective and objective method for estimating expert weights, combining subjective relevance with objective entropy-based measures.
• Proposed a prospect and regret theory-based TOPSIS approach for ranking decision-making options.

Main Results:

• The proposed method successfully selected an optimal AI tool for human-AI collaboration in a case study.
• Demonstrated the feasibility and stability of the developed model through comparative analysis.
• Showcased the superiority of the proposed model over alternative methodologies in handling uncertain decision-making.

Conclusions:

• The integrated approach enhances human-AI collaboration by effectively managing uncertainty and imprecision.
• The interactive consensus mechanism and novel expert weighting improve decision-making processes.
• The proposed LZN-based MCGDM framework provides a robust and superior solution for complex decision problems in AI.