PulseExploreJournal ClubDebatesTrendingResearchersJournals
Instagram
HomeExploreJournal ClubTrending
Synapse
⌘+K
Synapse
February 19, 2026Sustainability0 citationsOpen Access

Generative AI-Enabled Precision Recommendation for Green Products: Mechanisms of Consumer Cognitive Fluency and Low-Carbon Purchase Decisions

View Full Paper
KSKai SiCWCenpeng WangSWSizheng Wei

Key Points

  • This study aims to develop a GenAI-based recommendation system that enhances cognitive fluency and promotes low-carbon purchase decisions in green product markets.
  • Developed the Eco-GenRec system using large language models, multimodal generation, and retrieval-augmented generation techniques.
  • Conducted an experiment on a web-based shopping platform with 1000 participants in randomized control and experimental groups.
  • Measured cognitive fluency and low-carbon purchase conversion rates as primary outcomes.
  • Eco-GenRec group achieved a cognitive fluency score of M = 5.68 compared to control group M = 4.60, a 23.4% increase.
  • Low-carbon purchase conversion rate in the Eco-GenRec group was 36.3%, significantly higher than 17.6% in the control group, showing an absolute increase of 18.7%.
  • With high cognitive-style matching, conversion rate improvement reached 27.2%.

Abstract

To address the information-processing burden faced by consumers in green consumption markets due to complex carbon footprint labels, opaque certification standards, and vague descriptions of environmental benefits, this study proposes a generative artificial intelligence (GenAI)-based precision recommendation mechanism for green products. The mechanism aims to enhance cognitive fluency and promote low-carbon purchase decisions. An experimental system, termed Eco-GenRec, is developed by integrating large language models (LLMs), multimodal generation, and retrieval-augmented generation (RAG) techniques to enable personalized presentation of green product information. Based on inferred user cognitive styles, the system transforms product information into chart-based representations for analytical users or emotionally framed scenario narratives for intuitive users. This study is conducted on a web-based simulated shopping platform and employs a fully randomized design. A total of 1000 participants are randomly assigned to either a standardized information display group (control group) or an Eco-GenRec-generated display group (experimental group). Participants are drawn from diverse socioeconomic backgrounds and cover a wide age range. The sample exhibits substantial demographic diversity, which enhances the representativeness of the findings. Cognitive fluency and low-carbon purchase conversion rates are measured as the primary outcomes. The results show that the Eco-GenRec group achieves a significantly higher cognitive fluency score (M = 5.68, SD = 0.89) than the control group (M = 4.60, SD = 1.01). This represents an increase of 23.4% (t = 18.34, p < 0.001, effect size d = 1.17). In addition, the low-carbon purchase conversion rate in the experimental group (36.3%) is significantly higher than that in the control group (17.6%). The absolute increase of 18.7% is statistically significant (χ2 = 70.28, p < 0.001, effect size Cramér’s V = 0.265). Under conditions of high cognitive-style matching, the conversion rate improvement reaches 27.2%. Mechanism analysis shows that cognitive fluency mediates the relationship between GenAI-based recommendations and purchase intention. By transforming abstract environmental parameters into intuitive and easily interpretable content, artificial intelligence reduces information-processing burden and activates positive affect and trust among consumers. Overall, this study empirically validates the effectiveness of GenAI in green product recommendation. It provides a practical pathway for addressing the “comprehension barrier” in green consumption and extends the theoretical boundaries of research on cognitive fluency and low-carbon decision-making.

Ask AI
Helpful
Bookmark
Share
View Full Paper

Cite This Study

Si et al. (2026) studied this question.

synapsesocial.com/papers/6996a7a5ecb39a600b3ed8b4https://doi.org/10.3390/su18042018
Ask AI
Helpful
Bookmark
Share
View Full Paper