Insight
The Guru Gap: How Your Toughest AI Critics Can Become Your Most Valuable Champions
Organizations do not close the gap between AI experimentation and operational value by sidelining their skeptics. They close it by making their most experienced people co-designers of the solution.
Artificial intelligence is moving quickly. Organizations are deploying copilots, assistants and pilots across functions, and measure success of these tools by their adoption: who is using them, how often and for what tasks.
But in risk-sensitive, asset-intensive and highly regulated environments, usage alone is the wrong measure of progress. The real question is whether AI is helping people make better, more reliable decisions—without compromising safety, compliance, asset integrity or operational accountability.
Matt Mowrer
Sr. Vice President
Jake Stenzler
Chief Operating Officer
This is where the Guru Gap emerges.
The “gurus” are the deeply experienced engineers, inspectors, reliability specialists, risk professionals and subject-matter experts who know the domain, its requirements and its failure modes. They know what good looks like, and, just as importantly, they recognize when something looks plausible but is wrong.
These experts are often among the most skeptical users of generative AI. That skepticism is not a barrier to overcome; it is a critical capability to engage.
Understanding the Guru Gap:
Why Experts Are Skeptical
Many experienced professionals have already tested general-purpose AI tools. They have seen confident answers that are inaccurate, unsupported or disconnected from the specific rules, procedures and operational context that govern their work. In compliance and risk assessment, a hallucination is not merely an inconvenience. It can create a flawed decision, missed requirement, safety exposure or a compliance concern.
Their caution is therefore grounded in experience. This experience means they ask questions that may seem inconvenient or obstructive to AI champions but which need to be asked:
- What information is the model using?
- Can the answer be traced to a trusted source?
- What happens when the system is uncertain or wrong?
- Where should human judgment remain decisive?
- How will the organization validate performance as conditions, regulations and data change?
Those questions should shape the AI program, not slow it down.
The Cost of Leaving Gurus Out
When organizations treat experienced experts as AI resisters, they create a serious design problem. Less-experienced users may be enthusiastic early adopters, but they may not have the domain knowledge to distinguish a compelling output from a correct one. They may not recognize missing conditions, incorrect assumptions or exceptions that matter in the real world.
The result can be widespread experimentation without dependable value: generic prompts, disconnected tools and workflows that create more review work than they remove. In the worst cases, organizations generate what might be called “AI slop”: outputs that sound credible but lack the grounding, context and verification needed for consequential work.
That is not an argument against experimentation. It is an argument for putting the right expertise around it.
When Gurus Lean In
The most effective AI initiatives bring the gurus into the work early and give them a meaningful role in shaping it. This is especially powerful when AI models and agents are connected to curated knowledge layers: approved documents, rules, regulations and standards, procedures, engineering records, and relevant enterprise data.
With those foundations in place, experts can help define the questions the system should answer, the sources it should rely on, the evidence it must return and the circumstances in which it must escalate to a person. They can develop truth sets to evaluate accuracy, identify edge cases, refine prompts and retrieval patterns, and design workflows that use AI to support, not replace, professional judgment.
Their contribution does not stop at validation. Domain experts can identify the highest-value use cases because they know where work is repetitive, where decisions are difficult, where information is fragmented and where an error carries real consequences. They can help tailor agents and agent loops for activities such as compliance review, risk assessment, maintenance planning and reliability analysis.
This creates a productive feedback cycle: experts improve the AI system; the system reduces low-value effort and makes relevant knowledge easier to access; and experts can focus more time on exceptions, judgment and continuous improvement.
We have seen skeptics become some of the strongest AI users in their organizations once they can see how the technology works, what it can reliably do and how their expertise materially improves the result. Many would not have expected that outcome a year earlier.
Closing the Guru Gap
Leaders should not measure AI maturity by adoption alone. They should measure whether their best experts are actively helping define and govern the tools used in consequential work.
A practical starting point is to involve gurus in four areas:
Use-Case Design
Identify the decisions and workflows where AI can create value without exceeding acceptable risk.
Knowledge and Retrieval
Define the curated sources, priorities and evidence standards that ground responses.
Evaluation
Build truth sets, test edge cases and establish thresholds for accuracy, traceability and escalation.
Workflow Design
Determine where AI assists, where people approve and where automation is appropriate.
Closing the Guru Gap is not about convincing experts to trust every AI output. It is about building AI systems worthy of their trust and making their knowledge central to how those systems are designed, tested and improved.
The organizations that do this will not simply deploy more AI. They will deploy AI that is safer, more useful and more likely to scale into durable business value.
Helping You Go Further, Faster
ABS Consulting's AI Consulting Services safely unlock the value of an organization's data with custom AI solutions tailored to assets and operations, helping organizations improve operational efficiency, enhance safety outcomes and reduce operational risk.
