The growing divide between surface-level AI adoption and meaningful impact across safety-critical industries

Artificial intelligence has rapidly moved from experimentation to mainstream business adoption. Around 78% of organizations reported using AI in 2025, reflecting how quickly the technology has become embedded across business operations. Yet, only 25% of organizations said most of their AI initiatives achieved their expected return on investment, highlighting a growing gap between adoption and measurable impact. Against that backdrop, Dan MacDonald, founder and CEO of BIS Safety Software, believes many organizations assume they are embracing AI in meaningful ways when, in reality, they are only scratching the surface.

MacDonald observes what he explains as a growing divide between organizations that are integrating AI deeply into their operations, those that have largely ignored it, and a third group that believes it is transforming their business despite seeing little measurable impact.

His concern is particularly relevant in safety-critical industries where operational efficiency and workforce performance carry significant consequences. MacDonald believes this distinction is becoming increasingly visible across industries as organizations accelerate AI adoption while struggling to translate that investment into measurable operational outcomes.

According to him, surface-level adoption often looks productive at first glance. Teams may use AI tools to draft content, summarize information, or assist with routine tasks. While these activities can create incremental efficiencies, he argues that they rarely alter how a business fundamentally operates. In his view, meaningful transformation occurs when AI becomes embedded within workflows, automates complex processes, and enables work that would otherwise be impractical at scale.

“The difference is not the tool itself,” MacDonald says. “The difference is whether it creates a material impact on the business. Small efficiencies are valuable, but transformational deployment changes how work gets done across the organization.”

The gap often exists because leaders face competing priorities. MacDonald notes that many executives and managers spend their days solving operational problems and managing growth. As a result, they may not have sufficient time to explore how quickly AI capabilities are evolving or how deeply these tools can be integrated into business processes. From his perspective, many assessments of AI are based on experiences from months ago, despite the technology advancing at an extraordinary pace.

There are also cultural factors at play. MacDonald believes some employees may worry that improving AI-driven efficiency could reduce their long-term value to the organization. Others simply underestimate how significantly workflows can be redesigned. He argues that organizations often benefit most when experienced professionals combine their expertise with AI capabilities rather than viewing the technology as a replacement for human knowledge.

The consequences of misunderstanding this distinction can be significant. According to MacDonald, organizations are increasingly focused on achieving measurable returns rather than simply experimenting with AI initiatives. He believes that organizations should evaluate AI based on operational outcomes rather than adoption statistics alone.

Within BIS Safety Software, a company that develops training, compliance, and safety management solutions for various industries, MacDonald points to examples where AI is being integrated directly into daily work. One example involves voice-driven form completion technology that allows workers to have a natural conversation with the system while information is automatically organized into required documentation. According to MacDonald, internal testing showed significant reductions in form-completion time.

Another application involves course development. MacDonald explains that training programs that previously required weeks or months of development can now be generated and reviewed in a fraction of that time using AI-assisted workflows. He believes this capability could play an important role in preserving institutional knowledge as experienced workers retire and new employees enter the workforce.

“Organizations should be asking whether AI is creating measurable operational value,” MacDonald says. “If the answer is unclear, there is probably more work to do.”

As AI adoption continues to accelerate, the distinction between experimentation and transformation is becoming increasingly important. From MacDonald’s perspective, the future will be shaped by organizations that move beyond surface-level usage and focus on integrating AI into the core processes that drive productivity, learning, and operational performance. “Those conversations are no longer about whether AI is being used,” he says. “They are about whether it is changing outcomes in a meaningful way.“