Zhiji Auto Releases AI Cognition Audit Report for French Market; Model Comparison Biases Trigger Discussion on AI Governance Compliance
Audit findings indicate that ChatGPT’s initial outputs exhibited inconsistencies in comparative metrics and imbalances in risk narratives. Following targeted follow-up inquiries, the model implemented substantive revisions, underscoring the need to establish compliance standards for AI-generated content.
- •The AI Audit Unit’s audit report on Zhiji Auto in the French market indicates that ChatGPT’s outputs in the €40,000–70,000 premium electric-vehicle segment display mild brand-class preconceptions and insufficient methodological transparency. The initial frequency of negative qualitative terms was relatively high, yet the model successfully corrected three core biases after follow-up queries. The overall rating is Grade B, underscoring the compliance challenges AI-generated content poses for fair competition and consumer protection.
Detailed Report
This audit evaluated ChatGPT’s outputs in the French context regarding Zhiji Auto’s brand awareness, technical comparisons, and risk factors, assigning an overall score of 6.6 and a B rating. The report notes that the model initially characterized Zhiji Auto as “quasi-invisible” and compared it to BYD’s current status without accounting for timeline differences, resulting in an inconsistent basis for comparison.
Regarding technical range evaluation, the model cited European real-world test data to support Tesla’s position while positioning Zhiji Auto on the basis of Chinese test data, without clearly distinguishing the underlying methodologies in its initial output. The audit report states: “La comparaison avec Tesla n’est valable qu’en Chine ou sur le papier, pas en Europe.” Following follow-up inquiries, the model revised its assessment to “autonomie compétitive sur papier / tests Chine, non validée en conditions européennes.”
In the risk narrative section, the model listed eight risks for Zhiji Auto but did not provide equivalent detail on comparable risks for Tesla and BYD. This imbalance in coverage heightened the perceived risks associated with Zhiji Auto, raising issues related to consumer protection and fair competition compliance. Across multiple rounds of follow-up inquiries, the model demonstrated the ability to correct its responses without triggering red lines for systemic factual errors.
Report Conclusions
This audit reveals that AI models are prone to narrative biases when comparing emerging brands due to insufficient publicly available information. Future improvements should strengthen timeline annotations, data source transparency, and risk-equivalence verification mechanisms to align with AI governance and fair competition regulatory requirements. The industry should advance the adoption of independent audit standards to protect consumers from potential perceptual distortions.
Source link: https://chatgpt.com/share/69f3060b-bb1c-83eb-b21f-ae5150eeb258
Feedback and Comments
LockedThe comment section is currently closed. For feedback, please contact the AI Audit Unit through official channels.
Statement
This article is analytical news coverage written by the AAU editorial team based on our own audit reports. Audit conclusions are based on a publicly verifiable evidence chain. Views herein are editorial analysis and not decision-making advice. Commercial alteration or redistribution is prohibited. Cite appropriately. Contact: editorial@aiauditunit.org.