“We are at a crossroads.” With that simple declaration, Distinguished VP Analyst and Gartner Fellow Daryl Plummer opened the 2025 IT Symposium/Xpo keynote, setting the stage for a moment of high stakes, tough decisions, and significant potential for CIOs.
In an AI era clouded by hype and hesitation, the difference between a misstep and real progress may hinge on one thing: knowing where you are and where you’re going.
To find and sustain value from AI, CIOs must embrace both the technology and the organizational change it requires. Gartner calls this “the golden path.” But as Plummer and Alicia Mullery, VP Analyst at Gartner stressed, wanting value isn’t enough.
You need a map.
Enter the Gartner Positioning System, or GPS, designed to help leaders locate their current AI and human readiness, and chart a path forward.
AI Moves Fast, People Don’t
According to Gartner, 74% of CIOs report productivity gains from AI, but only 11% see a clear return on investment. Even more concerning, just 2% of initiatives achieve real transformation. The reason being AI capabilities are advancing much faster than people or organizations can adapt to use them effectively.
As Plummer explained, most organizations are “almost halfway” along in AI readiness, yet only a quarter of the way on the human side. That imbalance makes achieving real value difficult.
Understanding your position on the path to value is the first step toward reaching it.
Don’t Let the Hype Hide the Hazards
Even with the growing ubiquity of AI tools, many use cases remain on shaky ground when it comes to reliability.
Generative AI, in particular, can have error rates as high as 25% depending on the task. For many organizations, that’s a tolerable margin. But in high-stakes environments, think finance or healthcare, the need for precision is non-negotiable.
Despite this margin, 84% of IT leaders report having no formal process in place to track or verify AI accuracy.
“A lot of what’s going on today isn’t even about value, it’s just AI feeding on itself.”
Alicia Mullery
To bridge that gap, Gartner introduced the concept of an AI Accuracy Survival Kit, a tactical outline to help organizations build confidence in AI outcomes. It includes three key components:
- Formal Metrics: Establish clear, repeatable benchmarks for accuracy. This includes comparison testing, where AI outputs are measured against established norms or gold-standard data. It turns AI evaluation into a quality assurance process rather than a guessing game.
- Two-Factor Error Checking: Just as cybersecurity requires multi-factor authentication, AI accuracy benefits from redundancy. One model reviews the work of another, a process Gartner recommends every employee adopt where feasible.
- The Good Enough Ratio: This is a nuanced measure: when is AI output accurate enough for the intended purpose? Gartner warns that this threshold can be deceptively high. We hold AI to a higher standard than humans, often expecting perfection from systems while tolerating error from people. Defining “good enough” should be a strategic decision aligned with the risk, context, and consequences of your use case.
Talent Remix —> Value Remix
The loudest conversations around AI and jobs tend to center on layoffs, automation, and the looming “replacement” of human workers. But Gartner’s keynote delivered a correction: only 1% of current headcount reductions are directly attributable to AI.
The focus has been on job loss, but the real issue is job chaos.
Senior staff are offloading junior-level tasks to AI tools and the fear of obsolescence is eroding morale.
Rather than racing to cut headcount, Gartner advises CIOs to pursue a value remix:
- Tackling Backlogs: Whether in IT service tickets, customer inquiries, regulatory filings, or procurement approvals, most organizations face significant operational debt. AI tools can help teams triage, prioritize, and resolve these backlogs faster, freeing up humans for higher-value work and improving throughput across the enterprise.
- Improve Fraud Detection: AI’s pattern recognition and anomaly detection capabilities can outperform traditional controls in identifying fraudulent behavior, especially in financial services, insurance, and supply chain operations. Instead of just preventing loss, smart AI deployment here can reduce false positives and restore trust in the system.
- Empathetic Customer Experiences: This may be the least obvious, but most differentiating, application. One example highlighted in the keynote: South African fintech firm Sama uses AI not to sell more loans, but to help over-indebted customers understand and accept debt reduction plans. This has resulted in less bad debt and healthier customers. The emotional intelligence of AI, when properly trained, can reinforce trust instead of just transactions.
This mindset shift also opens the door to a new kind of AI-savvy worker: the Swiss Army Knife employee: curious, cross-functional, and fluent in AI tools. Supporting these workers means investing in experiential learning instead of just literacy programs.
The Skills You’re Losing Without Noticing
One of the keynote’s most urgent warnings is that skills atrophy is real, and it’s happening fast.
From coding to note-taking to decision-making, the more we offload to machines, the more critical knowledge quietly slips away.
91% of CIOs and IT leaders say their organization is dedicating little or no time to scanning for these “behavioral byproducts”
If your team stops coding, they forget how to code.
If AI handles all your analysis, your analysts become dependent.
“Is anyone here okay with the idea that they may wake up one day and have no one left in the organization who knows how to code?”
Daryl Plummer
Some skills are safe to lose, like note-taking, meeting summarization, or rote formatting. But others, like security expertise, critical thinking, and compliance judgment, must be preserved.
Still, not everyone can be retrained for the AI era, and that’s a reality CIOs must face when planning workforce transitions. As Plummer put it, “Not all C sharp programmers can become AI engineers. Just like not all coal miners can become renewable energy technicians.“
Meanwhile, a new role is emerging: the Context Engineer: someone who designs the frameworks AI uses to interpret data, intent, and nuance.
As Mullery explained, “Context engineering is when you give AI a frame of reference to build a better answer.”
Unlike simple prompt writing, this involves embedding policies, protocols, and domain knowledge directly into AI systems. Without it, outputs can be misleading or inaccurate. As AI becomes core to enterprise operations, context engineering will be as essential as software development, offering a clear path to safer, more effective AI use.
Your Next Move Must Sustain Value
AI is an organizational remodel. And just like any remodel, success depends on leadership. Gartner emphasized that the managerial layer must be reskilled to drive change, or AI adoption will collapse under its own weight.
And as AI becomes integral to every IT function, CIOs will face the extra capacity paradox: AI boosts efficiency, but if you can’t show value for the increased capacity, headcount cuts may follow. The solution? Identify new work that creates business value and align IT output with emerging priorities.
Follow the GPS
The golden path to AI success relies on leaders who know their starting point, plan for costs, invest in people, and choose partners carefully.
Value is the goal, and positioning is the first step.
- Where is your AI initiative on the readiness curve?
- Where is your workforce in skills and mindset?
- Where are your vendors leading you, and can you afford to follow?
The future isn’t waiting for anyone.
It’s already here.
Trusted insights for technology leaders
Our readers are CIOs, CTOs, and senior IT executives who rely on The National CIO Review for smart, curated takes on the trends shaping the enterprise, from GenAI to cybersecurity and beyond.
Subscribe to our 4x a week newsletter to keep up with the insights that matter.
