Tong, Ashley, Noor, Nurhafihz, and Zainol, Zahirah (2025) Generative AI Performance with DeepSeek: Key Service Quality Attributes from Users. In: Proceedings of the IEEE International Conference on Recent Advances in Systems Science and Engineering. From: RASSE 2025: IEEE International Conference on Recent Advances in Systems Science and Engineering, 4-7 November 2025, Singapore.

PDF (Author Accepted Version) - Accepted Version Restricted to Repository staff only until 31 December 2027.

Abstract

In the rapidly evolving Generative AI (GenAI) consumer market, DeepSeek, a Chinese artificial intelligence (AI) startup, has emerged as a disruptive force, captured global attention in early 2025 by leading app download rankings and triggering a notable decline in US tech stocks. Amidst widespread media coverage and comparisons against established competitors like ChatGPT, the specific performance attributes of GenAI that matter to users remain underexplored. This study addresses the gap by identifying the key service quality attributes of GenAI that are valued by consumers using DeepSeek. Content analysis on 953 DeepSeek user comments was done using Leximancer to reveal 11 key themes: “AI”, “better”, “app”, “voice”, “feature”, “China”, “using”, “OpenAI”, “accurate”, “update”, and “perfect”. The findings reveal that users highly prioritize attributes such as conversational interaction, multilingual capabilities, and information accuracy, which collectively shape their perception of service quality. Notably, the theme “better” highlights users’ comparative evaluations of DeepSeek against rivals like ChatGPT and Gemini, emphasizing its open-source nature and cost-effectiveness. Meanwhile, themes like “voice” and “feature” underscore the demand for multimodal functionality and continuous innovation, while “China” reflects geopolitical and data privacy considerations influencing user trust. The study extends service-quality perspectives to GenAI by articulating construct-level implications (e.g., multimodality, data-governance trust, comparative value) and translating them into concrete design and governance actions for practitioners. It extends beyond traditional frameworks like SERVQUAL by identifying a set of dimensions unique to AI-driven environments. For practitioners, the findings provide actionable insights for developers and managers to enhance user experience, prioritize feature updates, and address privacy concerns. By aligning GenAI development with consumer expectations, this research supports the creation of more competitive and user-centric AI applications. Future research directions include cross-platform comparisons and longitudinal studies to track evolving service quality standards in the dynamic GenAI landscape.

Item ID:	90208
Item Type:	Conference Item (Research - E1)
ISBN:	979-8-3315-4435-5
Keywords:	Generative AI, DeepSeek, Service Quality, User Experience, Content Analysis
Copyright Information:	© 2025 IEEE. Personal use of this material is permitted. permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. IEEE policy provides that authors are free to follow funder public access mandates to post accepted articles in repositories. When posting in a repository, the IEEE embargo period is 24 months.
Date Deposited:	29 Jan 2026 02:52
FoR Codes:	46 INFORMATION AND COMPUTING SCIENCES > 4602 Artificial intelligence > 460299 Artificial intelligence not elsewhere classified @ 50% 35 COMMERCE, MANAGEMENT, TOURISM AND SERVICES > 3506 Marketing > 350611 Service marketing @ 50%
SEO Codes:	22 INFORMATION AND COMMUNICATION SERVICES > 2204 Information systems, technologies and services > 220403 Artificial intelligence @ 50% 28 EXPANDING KNOWLEDGE > 2801 Expanding knowledge > 280106 Expanding knowledge in commerce, management, tourism and services @ 50%
	More Statistics