Generative AI: 80% Organizational Adoption, but 1 in 3 Cannot Use for Core Tasks; 'Adoption Wall' Reality Revealed

Uluru BPO Co., Ltd. (Chuo-ku, Tokyo; President: Yuhei Okeyama), a wholly owned subsidiary of Uluru Co., Ltd. (Chuo-ku, Tokyo; Representative Director, President & CEO: Tomoya Hoshi), which aims to solve labor shortage issues with a unique model combining 'AI and human power', conducted a survey titled 'Survey on Actual Usage and Challenges After Generative AI Introduction' targeting 1,201 business people belonging to companies with 100 or more employees.

■Survey Background

In recent years, the introduction of generative AI in Japanese companies has been rapidly progressing. With the spread of generative AI tools such as ChatGPT, expectations for operational efficiency and productivity improvement are higher than ever.

Based on this situation, this survey was conducted targeting 1,201 business people to clarify the 'actual usage' and 'adoption barriers' after the introduction of generative AI.

*When citing or publishing the survey data from this press release, please include the survey name [Uluru BPO 'Survey on Actual Usage and Challenges After Generative AI Introduction'] and the source URL (https://www.uluru.biz/news/25154).

■Survey Summary

Among those whose companies recommend company-wide introduction and use of generative AI (1,029 people), only 64.3% use it on a daily basis [Result 1]

Over 80% of people use generative AI for work within their organization (80.9%) [Result 4]

On the other hand, the tasks for which organizations have achieved operational efficiency and productivity improvement through generative AI are dominated by general-purpose tasks such as 'document creation' (47.0%) and 'drafting emails' (40.3%), while the sense of achievement in core/specialized tasks like 'business process automation' (15.3%) and 'advanced customer service' (11.9%) is below 20% [Result 8]

The biggest challenge in organizational use of generative AI is 'it can be used for simple tasks but not for core tasks' (33.6%) [Result 9]

Furthermore, among those who answered 'no time to focus on utilization', the percentage of 'heavy users' was the highest at 44.7%, revealing that the more advanced the usage, the less capacity there is for improvement and expanding the scope of application [Result 10]

■Survey Results

[Result 1] Analyzing the relationship between the status of generative AI introduction/permission in the workplace and the frequency of individual use, among the 1,029 people whose companies recommend company-wide introduction and use, only 64.3% answered 'use daily (almost every day)', and about 1 in 3 (31.8%) only 'use occasionally (a few times a month)'.

Although the environment for using generative AI is in place with company-wide introduction and recommendation, the fact that the total of those who use it infrequently or not at all exceeds 30% suggests that there is a significant 'wall' preventing daily use from taking root.

*This analysis is a cross-analysis of screening questions (both single-answer) regarding the status of generative AI tool introduction/permission in the workplace and the frequency of generative AI use in current work.

[Result 2] When asked about the generative AI tools currently used for work among the 1,029 people whose companies recommend company-wide introduction and use, 'ChatGPT (OpenAI)' was the most common at 55.2%, followed by 'Microsoft Copilot' at 50.3% and 'Google Gemini' at 47.7%.

On the other hand, 'Claude (Anthropic)' was only 12.3%, and 'In-house developed/customized generative AI' was only 9.4%.

While the evolution of Claude is a topic of discussion, the high adoption rates of around 50% for each of the three major tools indicate that an increasing number of business people are using multiple generative AI tools in parallel for work.

*From Result 2 onwards, the survey was conducted on the 1,029 people whose companies recommend company-wide introduction and use of generative AI.

[Result 3] When asked about individual usage of generative AI, combining 'use heavily' (38.2%) and 'use somewhat' (44.6%) gives 82.8%, revealing that over 80% of business people regularly incorporate generative AI into their work.

[Result 4] Regarding organizational usage of generative AI, the combined total of 'use heavily' (32.6%) and 'use somewhat' (48.3%) reached 80.9%, confirming that, similar to individual use, the use of generative AI is widely permeated at the organizational level.

[Result 5] When asked about initiatives for organizational generative AI utilization, 'organizing and structuring data into a format easily readable by generative AI (AI-Ready data preparation)' was the most common at 45.0%. This was followed by 'aggregation and digitization of 'knowledge' referenced by generative AI' (43.7%) and 'establishment of security environment and internal rules' (40.9%).

While prioritizing data infrastructure and governance, it appears that many companies are also concurrently advancing practical operational structures such as 'building a flow to verify information accuracy' (30.6%) and 'sharing prompts and creating templates' (26.9%).

[Result 6] When asked about current initiatives to promote organizational generative AI use among those who answered 'not using much / not using at all' regarding organizational usage, the combined total of 'actively promoting initiatives' (2.6%) and 'promoting initiatives' (28.4%) was only 31.0%.

On the other hand, combining 'working on it but not making much progress' (31.4%), 'working on it but not at all progressing' (12.2%), and 'unable to work on it' (25.3%) totals approximately 70%, indicating challenges in promoting generative AI utilization.

[Result 7] When asked about tasks for which individuals have achieved 'operational efficiency and productivity improvement' using generative AI, 'document creation using general-purpose generative AI chat (drafting emails, materials, etc.)' was the most common at 49.2%. This was followed by 'idea generation / text paraphrasing' (45.8%), 'drafting emails' (44.0%), and 'simple data entry/aggregation checks' (37.8%).

While many individuals achieve results in relatively easy tasks like document creation and editing, the sense of achievement in more advanced applications such as 'business process automation' (16.9%) and 'advanced customer service' (12.9%) is still limited.

[Result 8] When asked about tasks for which the organization has achieved 'operational efficiency and productivity improvement' using generative AI, similar to the individual results, 'document creation using general-purpose generative AI chat' topped the list at 47.0%. This was followed by 'idea generation / text paraphrasing' (39.3%) and 'drafting emails' (40.3%), indicating that utilization for document creation tasks is leading across the organization.

Both individuals and organizations are first experiencing the effects of productivity improvement from relatively low-hurdle document creation tasks, while the reality is that expansion into core and advanced tasks is yet to come.