What can digital transformation teach us about the current rush toward artificial intelligence? According to digital strategist and author Paul Boag, organizations rolling out AI today are falling into traps that feel less like calculated business strategy and more like a farce. Reviewers watching the corporate world’s handling of generative AI might easily dismiss it as unbelievable fiction, concluding that no real organization could possibly be that short-sighted.
Yet, businesses find themselves repeating the exact same errors made a decade ago. The frustration stems from a simple, inescapable reality: we have been here before.
Back in 2014, Boag published a book for Smashing Magazine titled Digital Adaptation, which examined the systemic mistakes organizations were making while attempting digital transformation and outlined what they should have been doing instead. Today, the similarities between that era and the dawn of the AI revolution are striking. Examining those past lessons offers critical insights for navigating the profound workplace challenges presented by artificial intelligence. Across both eras, the single greatest mistake is approaching a fundamental organizational shift purely as a technological challenge.
It’s About People, Not Technology
Undoubtedly, both digital transformation and artificial intelligence represent monumental technological advances. Some level of technical implementation is required in both cases, but that hardware and software deployment is merely the tip of a very large iceberg.
During the height of the digital transformation era, corporate IT departments busily deployed content management systems, customer relationship management software, and internal intranets. Simultaneously, customer-facing teams launched new websites, mobile applications, and automated personalization engines. However, organizations invested almost zero effort into understanding how everyday people were actually going to use these new systems. Internal software proved notoriously difficult to use, while customers either remained indifferent to what was being built or actively wanted something entirely different.
A nearly identical dynamic is playing out with artificial intelligence today. Senior management buys heavily into the market hype, leading IT departments to purchase enterprise licenses—such as widespread rollouts of Copilot Studio—and consider the job finished. Organizations might offer a brief, generic, company-wide training seminar, but that generally constitutes the entire strategy.
Just as before, everyday staff members are left wondering what they are supposed to do with these new systems. They feel mounting pressure to utilize the tools and deliver measurable business results, yet receive little practical guidance.
Organizations must recognize that artificial intelligence represents a cultural shift just as much as a technological one, and that profound cultural transformation carries enormous implications for their workforce. This time around, the stakes feel even higher for employees because they are genuinely fearful for their livelihoods. Staff members read the exact same alarming headlines as executives, and standard corporate promises from senior management do little to alleviate their underlying job security anxieties.
Why Would Staff Want To Adopt AI?
Examining the situation from an average staff member’s perspective reveals a stark truth: current AI deployments offer very few personal advantages, even setting aside the latent fear that management ultimately intends to use the technology to eliminate human roles.
The primary reason senior executives feel excited about artificial intelligence is the promise of boosted productivity. The technology allows employees to accomplish more work in less time. Yet, this efficiency does not inherently benefit the individual worker. Staff members are still expected to work their standard hours, but they now face expectations to produce vastly higher volumes of output within that exact same timeframe. They do not get extra time returned to their personal lives, nor do they receive higher compensation for their increased output. Under those conditions, employees naturally question why they should enthusiastically embrace the change.
Before asking staff to embrace artificial intelligence, organizations must first answer a critical question: how do they intend to genuinely motivate their workforce to do so?
One viable strategy involves helping employees identify the specific daily tasks they genuinely hate, and demonstrating how AI can help them complete those tedious chores more quickly and with less friction. People prove far more receptive to operational changes when leadership presents those shifts as genuine opportunities to improve their day-to-day working lives.
However, easing daily frustrations should serve merely as the baseline. Organizations may ultimately need to provide tangible incentives for AI adoption, ranging from clear career progression pathways and financial compensation to reduced working hours.
Staff Need Time And Support To Adopt AI
Adopting artificial intelligence is far from a trivial undertaking. It forces employees to challenge and dismantle nearly every established working practice they have spent years—or in some cases decades—developing. All of that accumulated procedural knowledge must be unlearned and rebuilt. That is an enormous demand to place on anyone under the best of circumstances, and the reality is that most modern employees are not operating under ideal conditions.
The typical worker is already overworked, chronically stressed, and struggling simply to stay on top of an overflowing daily to-do list. They lack the luxury of unscheduled time to sit down, patiently learn a complex new tool, and figure out how it naturally integrates into their existing workflow.
Adapting to these systems requires far more than casual exposure; it demands deep consideration and continuous experimentation. Sending an employee to a single standard training course and expecting them to harness AI to its full potential is entirely unrealistic.
At a minimum, staff require role-specific training tailored to their daily responsibilities. Beyond targeted educational programs, organizations must develop sophisticated, ongoing support mechanisms that allow employees across different departments to discover practical AI applications within their specific domains. Staff need protected time and space to experiment safely without penalty.
Formal training programs, however, possess inherent limits. In practice, the most effective support consistently comes from immediate colleagues. Rather than focusing initial rollout efforts on winning over the most resistant employees, organizations should identify the internal enthusiasts—the handful of people already experimenting with AI on their own initiative, often without formal executive sanction. Bringing these early adopters together, giving them the platform and time to share what actually works, allows organic enthusiasm to spread across the wider workforce.
This phenomenon was clearly visible during the digital transformation era. Internal conferences where staff demonstrated projects they built themselves did more to drive adoption than any top-down corporate memo ever could. People are far more likely to try a new approach when they see a peer in a similar role using it successfully, rather than when executives command them to do so.
Unfortunately, genuine experimentation comes with an intimidating downside: mistakes are inevitable.
It Needs To Be Safe To Fail
Many organizations maintain a pervasive culture of fear where professional mistakes are treated as undeniable signs of incompetence. Yet, the very nature of learning a sophisticated new technology like artificial intelligence involves trial and error, making a punitive internal culture a massive barrier to widespread adoption.
During the digital transformation era, senior management frequently demanded sweeping modernization while remaining genuinely confused when progress stalled. They failed to realize they were broadcasting mixed signals. While executives demanded digital innovation, they simultaneously continued enforcing rigid key performance indicators and penalizing any minor errors during annual performance reviews.
The environment offered no safe space to fail because the pressure to deliver flawless execution was relentless.
In the current AI landscape, those same toxic dynamics appear even more pronounced. Reports circulate of employees being sharply reprimanded for consuming too many token resources just weeks after leadership commanded them to adopt the technology. In other cases, workers receive such exhaustive lists of warnings regarding forbidden AI uses that they become paralyzed by anxiety, concluding that sticking to their traditional, manual methods is far safer than risking disciplinary action.
If management genuinely expects staff to adopt artificial intelligence, they must cultivate an environment where it feels safe to experiment. That requires adjusting rigid performance metrics, altering how staff performance is evaluated, and temporarily relaxing strict governance frameworks while employees are actively climbing the learning curve.
Handle Governance With Care
Naturally, artificial intelligence cannot be rolled out without establishing a reasonable baseline of governance. However, governance is another area that corporate entities have historically managed poorly.
During digital transformation efforts, internal governance typically swung to extremes: it was either completely nonexistent or so suffocatingly strict that it entirely crushed innovation. A remarkably similar pattern is emerging with current AI deployments.
Staff undeniably require a foundational framework of governance to understand their boundaries, clarifying what is expected of them and outlining clear parameters regarding permissible use. Yet, organizations frequently draft policies so restrictive that employees find it easier simply to abandon the tools altogether. Volumes of dense instructions detailing what data may or may not be shared with an AI model only amplify workplace anxiety.
Organizations would be far better served by establishing a concise set of core principles—guidelines short enough to fit on a few posters, comparable to established public sector design standards. Beyond those foundational rules, leadership must demonstrate a degree of institutional trust, allowing staff to apply their professional common sense.
Mistakes will inevitably occur, but those errors represent the unavoidable cost of organizational evolution. Management must accept that reality and build mitigation strategies directly into their rollout plans.
Adaptive Technological Rollout
Organizations must exercise equal caution when managing AI deployment from an IT infrastructure perspective. Once again, there is a powerful institutional temptation to lock down technological access completely, which inevitably stifles adoption.
Back in 2014, overstretched IT departments frequently selected a rigid suite of enterprise technologies and forced them organization-wide, offering little to no consultation with the actual end-users. Any software or tool outside that pre-approved list required navigating a complex approval labyrinth, usually involving a bureaucratic steering committee. The resulting glacial pace of change contributed heavily to the high failure rate of digital transformation initiatives.
Today, companies are repeating that exact blueprint by purchasing enterprise AI licenses, declaring the technology deployed, and ignoring the nuanced needs of specialized departments. For instance, specialized teams—such as user experience researchers—often find management pushing hard for AI adoption, only to discover that standard enterprise tools cannot support their advanced workflows. When teams find themselves bogged down in steering committee bureaucracy trying to secure alternative tools, their initial enthusiasm quickly wanes.
Faced with such internal roadblocks, many employees refuse to wait for official approvals. Just as "shadow IT" emerged during the digital transformation era—with workers secretly adopting external cloud storage tools because official corporate systems fell short—organizations are now witnessing widespread "shadow AI." When internal enterprise systems prevent employees from performing their jobs efficiently, workers simply open personal generative AI accounts on their mobile phones or home laptops to get the work done.
That workaround introduces significantly greater risks than the restrictions were designed to prevent. Sensitive corporate data ends up entering external tools completely outside organizational control, leaving no internal audit trail or record of what has been shared.
The instinctive corporate reaction is often to clamp down even harder with restrictive network policies, blocked websites, and warnings of disciplinary action. However, heavy-handed crackdowns rarely succeed; instead, they simply drive the behavior further underground. Shadow AI serves as a valuable diagnostic symptom, pinpointing precisely where official corporate toolsets are failing.
Rather than attempting to stamp out shadow AI through punitive measures, organizations should analyze it. By talking with staff about what they use personal accounts for and identifying the functional gaps in official tools, leadership can directly inform future technology rollouts.
While managing a chaotic, decentralized tech stack is far from ideal, early-stage technology adoption inevitably requires living through a degree of disorder while workers explore boundaries and discover common use cases. Organizations must permit staff to explore diverse applications and tools to determine what best suits their specific roles, tightening security controls later as genuine organizational requirements become clear.
A Dangerous Time
For many risk-averse organizations, a decentralized approach feels uncomfortably dangerous. Executives naturally prefer central control to ensure nothing goes wrong. Yet organizational change inherently carries risk, and in a rapidly evolving technological landscape, the risk of moving too slowly far outweighs the danger of moving too fast.
The most critical hazard facing modern enterprises is treating artificial intelligence purely as an IT installation project while overlooking the human element. AI adoption is not simply about deploying new software; just like the digital transformation era before it, success depends entirely on people, culture, and thoughtful change management. The sooner organizations recognize the true scope of that challenge, the better equipped they will be to navigate it.
