Michael Tyrimos is the Founder/CEO of Capacitor Partners, an operations and technology consultancy that transforms large-scale enterprises.
For more than a decade (and especially after the Covid-19 pandemic), ‘digital transformation’ was a dominant topic of discussion in almost every organization. Over the last couple of years, though, ‘AI transformation’ is overtaking the agenda, but often in a false manner. Many of the executives I talk to think that AI transformation is merely a subset of digital transformation, but it is actually way more than that.
Let’s take a step back first. What is digital transformation, exactly? It is not simply about moving work from paper to screen. It is when an organization starts to utilize digital tools in its everyday operations. Gregory Vial’s review of 282 works defines it as a process through which digital technologies alter how an organization creates value while prompting structural and organizational change.
Take an accountant, for example. Before the shift, they had a paper ledger, a calculator and filing cabinets. Afterward, they were in Excel and a cloud accounting system, with bank feeds coming in automatically and reports running themselves. A customer support agent had a phone, a paper script and a stack of policy binders before they were in a ticketing platform with macros, canned responses and a live customer history on screen.
The work became faster, more accurate and more efficient, but the underlying division of labor remained largely the same. People still performed the core tasks; digital tools primarily changed how those tasks were carried out.
What AI Transformation Adds
AI transformation, however, goes deeper. When a machine processes information, produces an analysis or recommends an action, the human role shifts from execution towards oversight, judgment and critical thinking. Employees must decide when to trust the output, when to challenge it and how to respond when it is wrong. AI therefore redefines roles, responsibilities and accountability.
Consider the accountant. In an AI-supported workflow, a model extracts invoice data, matches purchase orders, reviews vendor history and flags anomalies. The accountant may stop reading every invoice and concentrate on exceptions. This goes beyond faster processing. Employees must know when to trust the model, challenge its output and resolve unusual cases. Accountability remains human when execution is partly automated.
Customer support also shows this shift. Research by Erik Brynjolfsson, Danielle Li and Lindsey Raymond on 5,172 customer support agents found that AI assistance increased productivity by 15% on average. Less experienced workers improved the most, showing that AI can redistribute experience as well as automate tasks.
New Risks And Considerations
Digital and AI transformations share implementation risks, including integration failure, weak adoption and cybersecurity exposure. AI, however, goes beyond these to add new, ‘softer’ risks and considerations to the list. Let’s take a look at some instances.
Quality
Where must a person review, correct or take over from AI? Klarna illustrates why this boundary matters. In 2024, Klarna reported that its AI assistant independently handled 2.3 million conversations in its first month, equivalent to work by 700 full-time agents. Fast forward to 2025, its CEO acknowledged lower service quality, while the company recruited people so customers could reach a human. The lesson is to design human intervention into the service model and measure quality alongside speed and cost.
Accountability
Who is responsible when the output is wrong? If AI flags 12 suspicious invoices but misses two that later prove fraudulent, leaders need to know who set the threshold, who monitors performance, who can override the result and who owns the loss. Responsibility cannot be assigned vaguely to the model, the data team or the user. Every AI-supported process needs a named business owner and a clear route for human escalation.
Measurement
How will leaders know whether the redesigned work is improving? Cycle time, cost and adoption remain useful, but they are insufficient. Organizations must also monitor output quality by risk, human overrides, false results, escalation volume, complaints, model drift and performance across employee groups. Aggregate productivity can improve while quality declines in particular cases. Measurement must reveal where AI performs well and where human judgment still creates essential value.
Leadership: AI Transformation No Longer Belongs To The CIO Alone
Last but not least, both digital and AI transformation require executive ownership. The distinction is not that digital transformation belonged to IT while AI belongs to the wider leadership. Decisions about human authority, model autonomy, workforce design and acceptable risk cross organizational boundaries. The CEO should remain accountable for the outcome, while business executives own the processes they lead.
This direction is visible in BCG’s 2026 AI Radar, which surveyed 2,360 executives, including 640 CEOs. It found that nearly three-quarters of CEOs were their organization’s main AI decision-maker, twice the previous year’s share. This supports CEO ownership without removing functional responsibility. Technology leaders remain responsible for data, security and governance. Human resources leaders guide workforce redesign, while risk, legal and compliance leaders define controls.
The Key Takeaway
AI transformation creates a distinct organizational challenge. Indeed, both digital and AI transformation reshape how value is created and require leadership, new capabilities and redesigned processes. AI, however, reaches into cognitive work by producing analysis and influencing decisions. Leaders must therefore design the boundary between human and machine work, assign decision authority explicitly and retain accountability.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
