70% of Business Professionals Dissatisfied with Generative AI Output Accuracy — Usonar Survey of 500 Professionals Reveals Data Quality as Key to Improvement

Usonar Co., Ltd. (Headquarters: Shinjuku-ku, Tokyo; hereinafter "Usonar") announces the release of a survey report titled "Survey on the Actual Use and Output Quality Satisfaction of Generative AI," conducted among 500 business professionals currently using generative AI in their work. The report reveals that dissatisfaction with output accuracy reaches 70%, and a significant number of respondents spend unexpectedly long time verifying AI-generated content. To truly enhance productivity through generative AI, businesses must increasingly focus on high-quality data management. Download the full report for detailed insights. Survey Report on Generative AI Business Usage by Usonar Download the Report for Free Redirect to Usonar's Website ■ Background and Objectives of the Survey While generative AI adoption in business is expanding, users increasingly report frustrations such as "outputs cannot be used as-is" and "low accuracy requiring unexpectedly long verification time." The commonly cited cause is "prompt engineering skill." However, we hypothesized that beyond prompt skills, the "quality of data" referenced by AI might also be a critical issue, prompting this survey. - Quantify the "real-world situation" of business professionals using generative AI at work - Identify the root causes of dissatisfaction with AI response quality and reliability - Explore the relationship between data preparation and effective AI utilization ■ Survey Overview Survey Name: Survey on the Actual Use and Quality of Generative AI Method: Online survey (via Cross Marketing Inc.'s panel) Target: Business professionals who use generative AI "regularly" or "occasionally" for work Valid Responses: 500 Survey Period: March 2026 Conducted by: Usonar Co., Ltd. ■ Key Survey Findings (Summary) 1) About 70% express dissatisfaction with generative AI's "accuracy and reliability" (70.4%) 70.4% of respondents reported experiencing dissatisfaction with the accuracy and factual reliability of generative AI outputs, stating it occurs "frequently" or "occasionally." Despite widespread adoption, reliability remains a widespread concern. 2) Root cause of dissatisfaction is "data quality" — 61.4% attribute it to data When asked about the causes of dissatisfaction with AI responses, 23% cited "AI lacks specialized and up-to-date information," and 22% pointed to "potential inaccuracies in source data." Combined, 61.4% identified "data/information quality" as the root cause. In contrast, only 12.4% cited "insufficient prompt skills," indicating that most perceive the core issue lies not in user skills but in the data environment AI accesses. 3) Verification effort undermines productivity — 72.8% spend over 5 minutes on fact-checking and editing 72.8% of respondents spend more than 5 minutes per instance verifying AI-generated information. Some even spend over 30 minutes. The benefits of AI adoption are being offset by the time spent on output verification. 4) Key to advancing AI use is "data environment" — 61.6% identify it as most important When asked what is most needed to elevate generative AI usage, "improving data accuracy and freshness" (33%) and "access to broader factual data" (29%) ranked highest. Combined, 61.6% stated that "data improvement is most critical," surpassing "improving prompt skills" (20%). ■ Key Insight: The bottleneck in generative AI use is not "prompts" but "data environment" This survey shows that most users believe the primary cause of dissatisfaction with AI output lies not in user-level skills like prompting, but in the "accuracy," "timeliness," and "specialization" of the data AI references. Moving forward, businesses must prioritize creating environments that connect to reliable, accurate data to advance and improve AI utilization. ■ For Enterprise Data Quality Management, Choose Usonar Usonar provides enterprises with "LBC," Japan's largest corporate database. Corporate customer data, such as suppliers and sales targets, inevitably deteriorates over time due to data decay and input inconsistencies. Companies using Usonar maintain a continuous supply of high-definition business data, achieving a data environment with comprehensiveness, accuracy, and up-to-dateness. For details or materials, please contact us via our website. ※ About "LBC" LBC is Usonar's corporate database assigning an 11-digit management code to every business location nationwide. It centrally manages rich corporate attributes such as industry, revenue, profit, employee count, and fixed IP addresses. Matching client data with LBC enables precise targeting, excluding existing clients. Largest in Japan... Based on internal survey of Japanese corporate headquarters and business locations (as of 2026/6/1) https://usonar.co.jp/service/usonar/index.php#serviceLbc ＜About Usonar Co., Ltd.＞ Since its founding in 1990, Usonar has independently built and maintained Japan's largest corporate database, "LBC," to address B2B data challenges. Leveraging proprietary technologies for collecting and organizing high-quality, accurate data, along with cutting-edge expertise like artificial intelligence, Usonar provides the customer data integration solution "Usonar." We offer various cloud services to solve customer challenges, including the business strategy platform "PlanSona," corporate information and business card management app "mSona," and the registry information download tool "TodokeSona." Website URL: https://usonar.co.jp/service/usonar/ Company Name: Usonar Co., Ltd. Headquarters: 15F, Tokyo Opera City, 3-20-2 Nishi-Shinjuku, Shinjuku-ku, Tokyo Representatives: Chairman & CEO Nanami Fukutomi, President & CEO Katsuhito Nagata Founded: September 10, 1990 Listed: Tokyo Stock Exchange Growth Market, Securities Code 431A Business: Database marketing support services URL: https://usonar.co.jp/ Inquiries: Usonar PR Department (Hiyama, Egawa), Email: pr@usonar.co.jp