SURVEY: Generative AI Adoption Nearly Universal, Challenges Persist

A recent Bain & Company survey on generative AI readiness indicates that adoption among U.S. companies is nearly universal, even as organizations continue to grapple with significant roadblocks including security concerns, talent shortages and issues with output quality.

Why it matters: The swift and broad uptake of generative AI signifies a major shift in business operations and strategy. However, the path to realizing its full potential is complicated by challenges in effectively scaling the technology and integrating it securely.

The big picture: Generative AI has rapidly transitioned from an emerging technology to a common business tool.

• According to the survey, 95% of U.S. companies are now utilizing generative AI, an increase of 12 percentage points in just over a year.
• The depth of adoption is also increasing, with the average number of use cases in production doubling between October 2023 and December 2024. IT departments are seeing the fastest growth, though software code development remains the leading use case domain.
• While many applications focus on enhancing productivity and reducing costs, a significant number are also aimed at driving top-line growth.
• Reflecting this trend, AI has become a top priority for 15% of companies, up from 9% a year ago, and about half now report having clear implementation roadmaps.

Zoom in: Despite high adoption rates, companies face several persistent challenges.

• Data security and privacy concerns have intensified, particularly for firms that are more advanced in their generative AI implementation.
• A significant talent gap exists, with 75% of companies struggling to find the necessary in-house expertise for critical functions.
• The quality of AI-generated output remains a concern, though worries about accuracy are reportedly beginning to ease, suggesting increased confidence in the technology's capabilities.
• Companies in the early stages of adoption tend to be more concerned with organizational readiness, while those further along are more focused on data security, privacy, and the quality and accuracy of outputs, as well as vendor reliability.

Between the lines: The nature of challenges in generative AI adoption appears to evolve with experience. Pilot-stage companies often cite poor performance, data incompatibility, the need for process redesign, and securing leadership buy-in as primary frustrations. In contrast, companies with AI in production are more likely to point to issues with low-quality vendors, which can be difficult to identify early on. This suggests that as AI projects scale, new operational and external vendor-related pressures emerge.