A fuzzy multi-criteria decision-making framework for the comparative analysis of human and AI influencers.

The rise of artificial intelligence (AI) has led to a significant increase in the number of influencers in this field. This underlines the importance of understanding their influence and distinguishing them from human influencers. The existing literature lacks a systematic analysis of AI influencer performance and a mathematical weighting of evaluation criteria. This study addresses these gaps by pursuing two main goals: (1) determining the criteria that human influencers use to assess the effectiveness of posts and assigning weights to these criteria, and (2) ranking AI influencers based on these criteria from a human perspective. To this end, 12 expert human influencers defined 17 evaluation criteria. The study applied the multi-criteria decision-making methods Fuzzy Simple Weight Calculation (F-SIWEC) for weighting and Fuzzy Ranking Alternatives with Weights of Criterion (F-RAWEC) for performance ranking. F-SIWEC results reveal that "Ability to Convey Emotion" (C8) emerged as the most critical criterion with the highest weight (ω = 0.0643), followed by "Popular Topic Selection" (C15, ω = 0.0639) and "Voice Quality" (C4, ω = 0.0630), while "Background Selection" (C16, ω = 0.0535) ranked lowest, with overall weights ranging from 0.0535 to 0.0643 across the 17 criteria. F-RAWEC performance ranking identified Influencer 2 as the top-performing AI influencer (ρ = 0.2557), followed by Influencer 1 (ρ = 0.2031), Influencer 4 (ρ = 0.0114), and Influencer 3 (ρ = - 0.0105). Notably, AI influencers excelled in algorithm-driven criteria including "Popular Topic Selection" (average score: 4.561), "Content Description Text" (4.510), and "Animations/Effects" (4.391), yet scored lowest on "Ability to Convey Emotion" (2.906), the most highly weighted human criterion, revealing a fundamental "emotional gap" in AI's capacity to establish genuine human connections. Sensitivity analysis through rank reversal across four scenarios and comparative evaluation with five additional MCDM methods (F-MARCOS, F-WASPAS, F-SAW, F-MABAC, F-ARAS) confirmed the robustness and consistency of the rankings. This study is the first to determine the criteria and weights used by influencers and to apply F-SIWEC and F-RAWEC in evaluating AI influencer performance.