Learning Agility in the Age of AI: It Is Not About New Tools — It Is About Letting Go of Old Habits

When ChatGPT went live in December 2022, the immediate corporate reflex was predictable: open a training budget. In the first quarter of 2023, demand for ‘AI literacy’ courses on Turkish corporate learning platforms rose noticeably — and most of those programmes were designed as tool introductions. The result was equally predictable. Participants left those sessions and began using ChatGPT the same way they had always used Google: type a question, get an answer, move on. No change in decision-making. No real shift in how work was designed or reviewed. This pattern reveals something important about how learning agility is understood — and misunderstood — in most organisations. Learning agility is not the capacity to pick up a new tool. It is the capacity to abandon the habits and reflexes that made you successful — precisely when they still appear to be working. That distinction became one of the more consequential management questions of 2023.Consider a mid-sized insurance intermediary in Eskişehir, with 312 employees, holding its year-end leadership review in late 2022. When asked ‘Are you ready to use AI tools?’, every team leader raised a hand. When asked a different question — ‘Walk me through how you currently produce the analysis you give clients’ — the answers were uniform: Excel pivot tables, manually entered data, verbal summaries in weekly meetings. The willingness to learn a new tool was genuine. The willingness to examine whether the current workflow would actively resist integration was considerably lower. The real test of learning agility happens before the tool arrives: can you ask yourself what you will need to stop doing in order to make room for something different? The proportion of people who can answer that question honestly is consistently smaller than the proportion who can learn a new interface in a day.There are two competing views on this, and both are partly right. The first holds that learning agility is essentially dispositional — a relatively stable personality trait measurable at the point of hire, not reliably developed through training. The second argues that agility is a muscle: given the right environment and coaching, any leader can strengthen it meaningfully. The evidence from Turkey’s own recent experience is instructive here. During the forced remote-work transition of 2020 and 2021, SME managers who had previously resisted digital tools — cloud ERP, digital signatures, video collaboration platforms — adapted quickly under operational pressure. The crisis stripped away the ‘I cannot learn this’ defence. But when the acute pressure eased, a meaningful portion of those same managers drifted back toward older routines. This suggests agility is less a permanent acquisition than a state that requires ongoing conditions to sustain. It gets triggered; it does not automatically become permanent.In the Turkish SME context, the discussion has a structural dimension that the mainstream learning agility literature rarely addresses, because that literature is built on large organisations with functioning HR departments, visible career ladders, and role clarity. In most Turkish SMEs, decisions are concentrated at the owner or a two-to-three-person senior layer. Middle managers simultaneously run operations, try to think strategically, and treat any time spent in training as borrowed from somewhere else. Back in Eskişehir: in January 2023, one of the insurance firm’s team leaders was enrolled in a business intelligence workshop. She left halfway through because a client escalation came in. This was not a scheduling failure. It was a structural statement about what the organisation actually valued at that moment. In environments where learning time is not actively protected, the rhetoric of agility becomes exactly that — rhetoric. The shortest test of whether a company genuinely values learning: how many hours did a given employee spend, in the last quarter, on learning that was completely disconnected from daily operations? If that number is close to zero, agility is an aspiration, not a practice.Generative AI tools added a specific kind of pressure to this dynamic in 2023. Their most significant effect was not efficiency — it was cognitive discomfort. When a skilled analyst uses ChatGPT seriously for the first time, one of two things tends to happen. Either: ‘This makes my work easier.’ Or: ‘This can do part of what I do.’ The second reaction activates a set of well-documented defensive moves. Attention shifts toward the tool’s limitations — hallucinations, factual errors, the data security risks of submitting client information to an external model. These limitations are real and should not be dismissed. In Turkey, transferring personal data to servers outside the country requires either explicit consent or adequate protection safeguards under Article 9 of the KVKK personal data protection law — a genuine compliance tension that many companies quietly ignored when piloting ChatGPT in early 2023. For a firm handling insurance policy and customer data, feeding that information into a large language model is not a casual decision; it is a calculated risk with regulatory consequences. But here is the critical distinction: the difference between those who cite these limitations to close the conversation and those who understand them and still look for viable use cases is not a difference in technical knowledge. It is a difference in agility. The first group uses uncertainty as a reason to stop. The second uses it as a design constraint.Under what conditions does learning agility actually work, and when does it stall? Three conditions consistently matter. First: reducing the cost of letting go. The more a current method is bound up with status, peer approval, and established identity, the harder it is to change. A manager who has presented in the same format for a decade and received consistent praise for it needs more than instruction to try something different — the environment itself must signal that experimenting is valued, not merely tolerated. Second: psychologically safe failure. In Turkish business culture, tolerance for visible mistakes varies enormously across leaders. When the answer to ‘what happens if I try and it does not work?’ is genuinely unclear, people do not try. Third: making learning visible. When what someone learned, how they applied it, and what did not work can be shared openly within a team, learning becomes a social activity with compounding returns. The Eskişehir firm took a specific structural decision in March 2023: ninety minutes each week were placed outside the meeting calendar as a protected ‘learning corridor’ — no client calls, no operational escalations. Six months in, roughly 57 percent of team leaders were using that time actively. The remaining 43 percent still reported ‘not finding the opportunity.’ The opportunity had been created. The resistance was elsewhere, and that is precisely where the more important work needed to happen.The deepest trap in the current conversation about learning agility and AI is framing it as a tool problem. It is not. It is a question of who can let go of what, and when. ChatGPT became a trigger — a sudden, public, unavoidable signal that the rules of analytical work were shifting — but the trigger is not the test itself. Opening training budgets before examining what behaviours and structures block learning is like adding floors to a building without checking the foundation. The team leader in Eskişehir who made a habit of leaving learning sessions early was not blocked by a skill gap. She was blocked by a priority reflex that had been rewarded for years and had never been named as a problem. The real question for 2023 — uncomfortable but necessary — is how many people in that room could see that. And of those who could see it, how many were in a position to say so out loud.

This article was originally published in Turkish by Gökhan MERCANOĞLU on January 5, 2023. The English edition has been reviewed and edited by the author.


For dashboard management, the critical question is not which system to use. The real question is which problem will be solved, which data can be trusted, and which action will be accelerated. Without these answers, solutions look modern but only digitize old habits.


Gökhan Mercanoğlu
İş Zekâsı ve Raporlama