Bridging the AI Gap

One of the most marked trends in what’s being dubbed variously the “AI Divide” or the “AI Adoption Gap” is the friction between top-down implementation and roots-up adoption. There is a significant disparity in both the perception of and literacy in AI between executives and their larger workforce.

A 2025 Dayforce study finds that 87% of executives report using AI on the job, while only 27% of employees say the same.* This likely refers specifically to authorized AI tool use; as we will see in the next section, “shadow” or unsanctioned AI use boosts employee rates closer to 80%.¹⁶ Still, there exists clear divergence across hierarchies in perspective, confidence, and training.

While the vast majority of employees and executives believe that AI reskilling is pivotal, a contrast remains in terms of access. Dayforce finds that just 29% of workers have received AI training in the past year, compared with 81% of executives.^17,18 The Writer survey—which determines that AI is ‘tearing companies apart’—notes an average 30% disparity between executives and staff. For example, only 33% of employees feel their organization is AI-literate, while C-suite leaders cite high literacy at 64%. In fact, the very definition of success appears to be in dispute, with 75% of C-suite leaders asserting that their AI strategy has been “very successful” compared to just 45% of employees.

The Resistance Factor

These are among many findings to suggest that, while leadership observes impressive technical deployment and initial tools training, the teams assigned to use those tools find them disconnected from their iterated, daily workflows and established efficiency protocols.^18,19 As Greg Shove of Section AI puts it, “Everyone has access [to AI], almost no one is generating value.”²⁰

Findings from IBM Insights mirror this tension. Only 33% of employees report that their employers are fully realizing AI’s potential, and a jarring 30% of younger employees cite “leadership resistance” as the foremost barrier to full AI integration.¹⁶

Neither is this friction solely a matter of varying perspectives, as corporate AI implementation is being disrupted by staff themselves. Writer finds that:

41% of Millennial and Gen Z employees admit they’re sabotaging their company’s AI strategy, for example by refusing to use AI tools or outputs. The reasons for this range from fears about AI taking over their job to concerns about the quality of AI tools, pointing to the need for a better change management process.¹⁹

* That other, contemporary research suggests much higher employee usage is likely explained by specific description here of “authorized” AI tools, which is corroborated elsewhere at about ¼ adoption.

The Bottom Line

Ultimately, the current mode of AI implementation is not only failing to produce ROI, it is actively straining synergy and morale.

When AI is deployed from the top—however competently—without a corresponding cultural framework, it is experienced by the workforce not as a tool for empowerment, but as a source of organizational friction that both threatens to proceed without them and yet overlooks their road-tested, established practice.

While organizations are allocating billions toward technical AI infrastructure, current data suggests that, beyond “day-one” proprietary training, staff are not being sufficiently introduced to AI as a broad technology or as a steady fixture of daily workflows. Section AI notes in its 2026 AI Proficiency Report, “Employees who have undergone AI training score, on average, 40/100 in AI proficiency”¹⁸—suggesting that training is insufficiently value-centric.

A 2025 Microsoft study corroborates this disconnect, finding that nearly two-thirds of leaders acknowledge a pervasive AI literacy-skill gap. Crucially, the same study reveals that 48% of employees rank sufficient training as the most important factor for embracing company-led adoption, and that employees who feel sufficiently trained are nearly twice as likely to report realizations of AI value. For organizations who move past pilot programs to enterprise-wide literacy, the results are quantifiable: “90% report faster decision-making, 81% report increased revenue, and 81% report better employee retention.”²⁹

Structural Obstacles: Accessibility & Bandwidth

The core obstacles to necessary widespread literacy appear to be structural—specifically, accessibility and the “capacity to engage.”²⁹

Current training programs frequently prioritize high-visibility teams or departments (e.g., sales and marketing), leaving many employees excluded from a shared strategic vision—with significant risks to morale and organizational mobility. Furthermore, employees already burnt out on existing workloads lack the cognitive surplus required to experiment with new tools, and where employees are prepared to engage, company-endorsed schedules for learning and experimentation are lacking.

We might be reminded of Google’s original 70-20-10 Rule, which effectively earmarks 10% of resources for experiment, producing outsized reward and market-leading innovation. The effective inverse of this rule, as research into AI adoption indicates,²⁹ is that without dedicated time and resource-aware programming for exploration, competence and therefore market potential remain stagnant.

When comprehension is distributed throughout the enterprise, there are numerous benefits across hierarchies

Misalignment of Capital Allocation

Regarding accessibility, MIT research indicates a notable misallocation of resources: half of generative AI budgetary spending (and 70% of AI allocation overall) has been directed toward sales and marketing, despite evidence that ROI is significantly more impressive when expenditure focuses on back-office automation.¹

Such disparities underscore the value of a diversified top-down AI strategy, and why enterprise-wide literacy programs are key drivers of profit. When comprehension is distributed throughout the enterprise, there are numerous benefits across hierarchies, including two primary among front-line staff:

• Deep integration: Workers can more capably identify high-value integration points for AI within their own, real workflows.
• Mitigated resistance: AI literate employees are less likely to fear being replaced by AI and automation^30,31 and are, logically, less likely to sabotage implementation efforts.

These findings reflect core Ankora theses: that cultural strategy is integral to real adoption, and that epistemic cohesion—the alignment of perspectives and comprehensions of AI across a company—will determine leaders in this space.

Most firms struggle to capture real value from AI not because the technology fails—but because their people, processes, and politics do.

Feng Zhu, Harvard Business Review⁹

It is clear—when regarding the slim minority of corporations enjoying meaningful returns on AI investment—that technical implementation alone is not enough to achieve success. As challenges to integration consistently center on such themes as perception across hierarchy, disparities in trust and skill, and varying awareness (and existence) of governing protocols—in effect, culture—it is evident that organizational and cultural strategy is the chief missing component in most AI rollouts.

Organizations ideally should allocate 70% of their AI-initiative resources to supporting people and adapting human-centered business processes

This is endorsed by the findings of leading authorities who emphasize human-centered requisites for AI transformation. Boston Consulting Group (BCG) has notably applied Google’s 10-20-70 Rule to AI roadmaps. They conclude that in order to capture maximum value, organizations ideally should allocate 70% of their AI-initiative resources to supporting people and adapting human-centered business processes.⁴¹

Gallup recently underscored this in a report titled, rather frankly, “Your AI Strategy Will Fail Without a Culture That Supports It.” They argue that successful leaders must take the following human-centered actions:

• Diagnose Cultural Readiness: Utilize qualitative and quantitative assessments to inform the firm’s strategic roadmap.
• Align Investment with Purpose: Ensure AI deployment reinforces the organization’s "why" and core competitive differentiators.
• Communicate a Clear AI Narrative: Create an engaging "rational and emotional case" for employees. Gallup finds that when this plan is communicated clearly, employees are 2.9x more likely to feel prepared and 4.7x more likely to feel comfortable using AI in their roles.
• Sustain Adoption through Habit: Intentionally foster and replicate the cultural behaviors and success stories that turn initial enthusiasm into daily operational habits.

McKinsey likewise identifies cultural readiness as the primary indicator of cognitive-technology success, subtitling their own leading report, “Empowering People to Unlock AI’s Full Potential.” They remind readers that “3x more employees are using gen AI […] than leaders imagine” and yet that nearly half feel they are insufficiently supported for adoption. McKinsey’s conclusion, like many in the space, is that “this is the moment for leaders to set bold AI commitments and to meet employee needs with on-the-job training and human-centric development,” to co-create AI implementation strategy with employees “from the bottom up,” and to rewire a “culture of autonomy,” collaborative learning, and transparency within their organization.¹³

Culture is not just a prerequisite for AI success, but a core beneficiary of it

This sentiment is multidisciplinary and found at each scale of AI research. At the individual level, both neuropsychologist Stephen Kosslyn and the MIT Media Lab confidently declare that AI is problematic at best when used to replace human capacities, leading to consistent underperformance at “neural, linguistic, and behavioral levels.”²⁸

However, where AI tools are managed to augment such capacities, unprecedented achievement may follow. As Kosslyn puts it, AI is most successful as a “cognitive amplifier:” an enhancement rather than a replacement of human problem-solving.⁴² Whether discussed as a business tool or a personal implement, “human-in-the-loop” planning appears to be the emerging wisdom.

Crucially, AI success and cultural stability exist in a virtuous cycle. A joint MIT Sloan-BCG study found that, among organizations with successful AI implementations, more than 75% saw improvements in team morale, collaboration, and collective learning.⁴³ This reinforces that culture is not just a prerequisite for AI success, but a core beneficiary of it—representing overall a feedback loop of qualitative and quantitative gains.

This trend is repeatedly noted by a growing variety of trusted institutions—such as IBM, Deloitte, Gartner, and Harvard Business School—who each declare in so many words a growing consensus: where organizations introduce enterprise-wide AI literacy and cultural programs, otherwise stray returns may be captured.^12,44–46

Culture, it may evidently be said, is the soil in which AI strategies perish or thrive.

Naturally, the first step to integration and ROI is to determine the true uses, perspectives, and literacies currently available within an organization. Where global hype, false expertise, and majority opaque AI usage present massive distortions to the field (deserts of the real)—and where even companies without formal AI strategies are being transformed by shadow AI use—it is vital that firms establish a clear AI profile.

As Bloomberg Law proposes:

Taking adequate time and resources to properly evaluate can help protect businesses from financial losses and other risks.³⁴

The statistics identified in our ‘Challenges’ section, for example, suggest that idealism regarding AI strategy is a core driver of failing adoption. Or as BCG straightforwardly notes, “Teams with a more realistic view of the road ahead tend to make faster progress.”⁴⁷

Yet leaders may be induced by media hype and proponent claims to view investments through rose-colored glasses. Meanwhile, front-line workers are experiencing morale declines as they encounter day-to-day realities that highlight clear gaps between implementation theory and practice.^19,10

Companies […] must first gain clear-eyed, unbiased intelligence on how AI is genuinely perceived, understood, and applied throughout their organization

Forensic Discovery Levers

To avoid an AI ‘emperor-wears-no-clothes’ scenario, companies preparing implementations or seeking returns on established investments must first gain clear-eyed, unbiased intelligence on how AI is genuinely perceived, understood, and applied throughout their organization. Primary levers for this discovery include:

• Skills Assessments & Surveys: Identify the delta between presumed and proven perceptions, comprehensions, and applications.
• One-on-One Interviews: Capture the nuance of organizational culture—including granular detail on evolving trends, morale, and internal use cases—while identifying the underlying shifts required for strategic alignment.
• Statistical & Qualitative Analysis: Rigorously synthesize the survey and interview data, transforming raw feedback into thematic organizational facts.
• Actionable Reporting: Create a roadmap from the data to key decisions, leveraging existing organizational momentum to determine evidence-based paths to profitable integration—including competencies, governance, and key narratives.

With this diagnostic clarity, firms can move beyond generic implementation strategies to integrate high-precision AI frameworks that leverage the unique strengths of internal talent and existing operational logic—thereby securing a clear path to measurable returns and long-term brand equity.

Forensic AI discovery is not only a tool for determining how a company is uniquely positioned to profit from AI; it is also a vital defense against AI washing. Organizations lacking a full understanding of their AI profile are prone to overpromise capabilities under the intense pressure of a global innovation atmosphere.

As we have established, such reality-skewing effects can expose firms to a variety of damaging scenarios, including higher vulnerability to market corrections, loss of stakeholder trust, and high-priced regulatory fallout.

XAI & Forests of the Real

Time and again, researchers and advisors commend that the solution to any number of underlying causes of the AI Gap is to ground in genuine capacity and performance

There are, on the upside, a number of methods by which organizations can prevent or reverse AI washing trends. A primary means of capturing gains and developing the AI-washing inverse, explicable or “explainable AI (XAI),”⁴⁸ is to lead through what we have termed forests of the real. Time and again, researchers and advisors commend that the solution to any number of underlying causes of the AI Gap is to ground in genuine capacity and performance.

To invert patterns of AI washing and secure a position of market leadership, organizations must move beyond superficial signaling. The following compact framework—distilled from insights by California Management Review and Bloomberg Law—provides a roadmap for authentic, value-driven AI adoption:

• Audit-Backed Transparency: Demand "verifiable proof for AI claims” and deploy "rigorous internal audits"—giving particular attention to accurate language in public disclosures to ensure claims match reality.
• Value-Driven Policy: Shift from nominal short-term gains to genuine long-term value by establishing enforceable governance protocols and AI-specific KPIs that measure real outcomes (e.g. “ROI from AI initiatives or improvements in efficiency”).
• Literacy & Skeptical Innovation: Build organizational literacy and a culture of skeptical innovation, where leaders and employees are free to experiment and simultaneously to challenge project efficacy or demand proof.
• Prudent Leadership: Leaders can “reject the allure of inflated AI claims” and act as culture-strategy role models: exercising prudence with industry buzzwords, challenging or even walking back AI output, and engaging verified experts to ensure AI initiatives are accountable and defensible.^33,34

In sum, corporate leaders and their staff can generate durable value and avoid the risks of AI washing by establishing evidence-based insight, clear protocols, and grounded measurable implements.

“Be real” may be the simple if ironic mantra that precedes enviable success in the artificial-intelligence age.

A critical outcome of forensic AI auditing is the discovery of shadow AI—the reality of how an organization is utilizing unauthorized tools. Some 60% of employees use free external tools, a quarter pay out-of-pocket for off-premises subscriptions, and 28% admit that they would continue even if such tools were “outlawed” by management.²¹

Organizations must learn to illuminate shadow AI as a standard hygiene practice, identify and mitigate its risks, and merge it with top-down AI programs

As public AI tools continue to grow in marvel, efficiency, and human likeness—fostering parasocial bonding and enhanced loyalty^49–51,32—the disinclination to abandon off-prem tools will only accelerate. Organizations must learn to illuminate shadow AI as a standard hygiene practice, identify and mitigate its risks, and merge it with top-down AI programs. After all, where opaque public tool use is obviously risky, its risks are manageable with clear safety protocols and consequences, and its potential rewards are robust.

The Risk Profile

Once discovered, the risks of shadow AI use can be remedied by means tailored to company profiles. In general, policy must clarify which AI practices are of undue risk and establish well-communicated consequences for violations—as it is apparent that most shadow AI risks are the result of low literacy and unclear governance.^52,53 This applies to concerns including, chiefly:

• Data Integrity: Preventing breaches, safety risks, and sensitive disclosures to public platforms.
• Operational Oversight: Mitigating overreliance on AI tools and responses. This is addressed with human-in-the-loop strategy, a practice that is both protective and rewarding.⁵³

The Strategic Archive

The considerable potentials of shadow AI use may also be activated once diligently unearthed. It must be understood that, by and large, staff utilize shadow AI not to shirk responsibility or needlessly defy protocol, but as a means to improve productivity.^16,21

This presents an opportunity to unite theories of implementation with established practice. Organizations can:

• Condense rollout times by subsuming previously undiscovered progress.
• Recognize once-hidden innovation in workers who have already planned or iterated AI solutions.
• Upraise morale, staff loyalty, and internal synergy by welcoming staff contributions as emergent best-practice.

Ultimately, shadow AI is an archive of untapped talent, passion, and efficient application

Often, top-down AI programs are rejected by employees simply because they fail to account for the level of practical efficiency already employed in daily shadow workflows.^54,55 Ultimately, shadow AI is an archive of untapped talent, passion, and efficient application that can be de-risked and combined with formal initiatives.

These practices effectively indicate where AI adoption has naturally occurred—representing a locked value in terms of organic, company-specific R&D as well as a pivot point for organizational morale.

Enabling front-line staff to feel ‘seen,’ valued, and included in policy development is its own well-established gain.^55–58 Such cultural development simultaneously overlaps with leadership needs for deep insight and swift integration—a win-win scenario that provides increased protection and profit in the face of dynamic market trends. It also speaks to a further solution to AI implementation challenges.

Deep auditing and strategic merging with shadow AI are foundational to what we at Ankora describe as epistemic cohesion: the degree to which stakeholders across an organization demonstrate a shared vision, competency, and rule of law.

Successful companies, particularly under technological change, require a significant degree of unified perspective. While this may at times appear self-evident, researchers in a variety of fields have taken to illustrating the point. Peter Senge’s foundational work in organizational theory, The Fifth Discipline, names “shared vision” as a pillar of organizational success.⁵⁹ Researchers publishing in the Journal of Applied Psychology find that “shared mental models” significantly improve team performance,⁶⁰ while a 2022, peer-reviewed article in Empirical Software Engineering finds that “Coordinating mechanisms are needed to facilitate [teamwork effectiveness],” including “shared mental models, communication, and mutual trust.”⁶¹

To effectively root AI strategy within daily organizational culture, there must be a shared baseline for thinking about the technology

This requirement for epistemic cohesion is echoed across research into AI success specifically: from the National Academies’ evidence that "shared mental models" are the chief way for humans to effectively partner in AI,⁶² to McKinsey’s observation that “Employees tend to rally behind a shared vision of [AI] opportunity.”⁶³ The Defense Business Board’s 2024 report on Department of Defense efficiencies further grounds this sentiment, stating that: “Establishing a baseline for a common digital language, with clear expectations and data requirements, is a critical first step.”⁶⁴

To effectively root AI strategy within daily organizational culture, there must be a shared baseline for thinking about the technology. Naturally, individual stakeholders may likely and often should have differing opinions—as diversity of approach is a well-recognized key to success^65,66—yet ideas and pursuits must overlap at a common site.

Members of a world-class orchestra, for example, may expectedly differ in musical preference or default interpretations of a piece, yet harmonious performance can be achieved by, essentially, shared training (rehearsal), leadership (conducting), and point of reference (sheet music).

In the realm of corporate and organizational AI, epistemic cohesion requires that stakeholders:

• Align on Objectives: Reach a broad agreement on the ultimate goal of any given AI initiative.
• Establish Baseline Competency: Share a fundamental understanding of AI capabilities and limitations—with a unified sense of where internal expertise resides.
• Define Clear Protocols: Maintain a cohesive grasp of the rules and governance regarding AI.

Together, these elements transform a fragmented digital landscape into a singular, resilient operating environment. Such an environment requires that every member of the organization possesses the fluency necessary to interpret and execute its proposed shared vision.

Achieving this level of systemic alignment necessitates a move beyond initial technology exposure and toward a rigorous standard of AI literacy—as a necessary foundation on which institutional competency and trust may be constructed.

Once AI profiles are sufficiently mapped and assessed, a primary means of gaining epistemic coherence is to program literacy development. ‘Literacy’ is overlapping but distinct from ‘training’ in that the latter tends to imply a narrow focus on technical skills, while literacy describes a “holistic proficiency.”⁶⁷

As we at Ankora define it, literacy in AI is comprised of three vital components:

All employees are currently exposed to—and in positions to deploy—artificial intelligence

AI literacy programs need be modular, regular, “measured by impact not attendance,” and addressing of typical use cases within a given organization or firm.^7,68

While more than half of employees believe that upskilling is a pivotal need for successful AI adoption,¹⁷ many firms make the strategic errors of either treating AI learning as an exclusive privilege for technical or ‘AI-interfacing’ teams, or failing to provide support for staff to experiment with new technologies.

In reality, all employees are currently exposed to—and in positions to deploy—artificial intelligence, and thus benefit from the opportunity to gain in baseline competence through formal literacy programming and established time and resources to experiment. Meeting this broad need is a strong step toward simultaneously mitigating risk-governance concerns while fueling innovation and value capture.

To summarize, the strategic value of AI literacy is to be found in these key areas:

• Catalyzing Epistemic Cohesion: Literacy ensures that cross-department, technical and non-technical staff speak a common language, raising teamwork capacities and preventing silos that can lead to project failure.
• Building Market Resilience: A literate workforce acts as a filter against "snake oil" and AI washing. By recognizing the difference between stochastic parrotry and deep reasoning, nominal claims and real capabilities, a firm can merge grounded tools with agile practices to grow through market correction.
• Elevating Morale and Retention: Upskilling leads to greater organizational trust and confidence, reducing tech anxiety and initiative sabotage while raising felt sense of belonging and appreciation by employers.
• Enabling Decentralized Innovation: Literacy distributes innovation. When it is widespread, "AI champions" emerge from front lines, discovering high-value use cases that might otherwise remain invisible to leadership from the top down.

Companies that seek profitability in AI—and heed the persistent advice regarding human-centered integration—will gain much by establishing literacy programming as an essential process. If culture is the soil in which AI thrives or dies, literacy is a vital seed for a crop that will survive winter.

In the probable scenario that the market will see significant corrections regarding AI investment, those companies that manage to remain ‘under the wind’ will be those that invariably sidestep the large-scale seductions of AI washing and the various products and ideas on offer that may soon prove to be so much snake oil.

Technology adoption for its own sake has never created value

The Survival Traits of High-Value AI Business Models

The era of cognitive technologies demands foundational comprehension, shared organization-wide understanding, and implements rooted in daily workflows. Based on lessons of the dot-com boom, numerous experts suggest that companies succeeding in AI must, ironically, be real. To survive and gain from a likely natural selection event, business models will need these durable strategies and attributes:

McKinsey reminds corporations that “Technology adoption for its own sake has never created value.”¹⁰ If an AI bubble bursts, it will likely reflect the dot-com era form of trough, revival, and natural selection.⁷⁰ As Khurram Akhtar, writing for Forbes Business Council suggests: “Much like the tech giants that emerged stronger after the dot-com crash, the AI business models that survive will shape the next decade of digital transformation.”

The core characteristics experts cite as essential for survival—and for realizing the potential of subsequent outsized influence—are those anchored in genuine value. This means prioritizing pragmatic, in-workflow solutions that generate immediate utility for customers while operating within organizational strategies designed for grounded adaptability.

Ultimately, the strategic consensus is clear: When the wind blows, companies must have tended their roots.

It is projected that companies will spend more than $3 trillion on AI infrastructure by 2028,⁷² yet the vast majority of returns have yet to be realized. AI will inevitably transform the world of business and the world at large—though which of its many promises are fulfilled remain to be seen.

Just as entities grounded in the real logics of digital networks experienced outsized reward following the dot-com bubble, the winners of this era will be corporations that proceed from genuine, unbiased insight into their own AI profile. Organizations that cultivate such insight and deploy it through a fundamental, transferable comprehension of the technology—with applications specific to their business—will thrive where superficial ventures fail.

AI success begins by identifying the cultural disparities and locked values that currently obscure bottom-line risks and opportunities. By performing this essential stock-taking, leaders can build adaptive implementation programs that drive true integration and secure ahead-of-the-curve, lasting return on investment.

ASG is a strategic advisory firm identifying human-centered solutions for the cognitive technologies of the future