"Automating Uncritically With AI: The Mistake Companies Can't Afford"

When adoption becomes widespread, the key question is no longer what AI is capable of, but what happens when it is deployed at scale. The challenge is not a technical one; it is an organizational one. Companies that fail to rethink how work is designed and carried out risk getting it wrong.

AI is already delivering meaningful time savings. Tasks that once took hours (preparing documents, synthesizing information, running preliminary analyses, comparing scenarios) can now be completed in minutes. Productivity rises. But efficiency gains on their own do not guarantee sustainable value. When a technology reshapes how work gets done, it also reshapes how work should be organized. If that redesign isn’t approached deliberately, it gets shaped by default: poorly conceived automation, roles that gradually lose substance, and short-term decisions that undermine talent development over time.

This dynamic is already becoming visible in areas most exposed to automation. When AI takes on part of the early-stage work, it’s easy to assume that certain roles, especially junior ones, are no longer needed. That conclusion, however, is not only a weak approach to building capabilities; it also reflects an unrealistic view of the future of work. It’s possible to gain efficiency today and sacrifice resilience tomorrow.

A better way to frame the discussion is to recognize that humans and AI are not competing for the same work. AI excels at speed, consistency, synthesis, and repeatable execution. People continue to add the most value through judgment, context, negotiation, empathy, accountability, and responsibility. Real impact emerges when work is designed around complementarity, not broad replacement.

In complex functions such as risk, legal, strategy, or business, AI can help structure information, surface signals, and prepare scenarios. That’s already happening at BBVA. But decisions themselves, the judgment behind them, the accountability they carry, and the reasoning that supports them, remain human. AI certainly shifts work. But more importantly, it shifts it toward higher-value activities.

Which leads to a point that often gets missed: time saved does not automatically become value created. Time freed up by AI can be reinvested in quality, innovation, or better service. It can translate into greater productive capacity, or it can dissolve into more urgency, more meetings, and more low-value work. The difference isn’t set by the technology itself, but by the organizational choices that surround it.

That’s why the next step isn’t simply to use more AI, but to organize work more effectively around it. This is the reflection now underway across our organization: Where does automation genuinely add value? Where does collaboration between people and AI make more sense? And where is a high degree of human judgment essential? A truly mature conversation about AI doesn’t focus on automating roles, but on automating tasks. The distinction matters. Rather than talking about jobs disappearing, we should be talking about how groups of tasks are redistributed between people and intelligent systems, allowing roles to evolve instead of being stripped of substance. This shift in perspective fundamentally reshapes the conversation about employment and productivity.

There is also growing evidence that AI’s impact will not be the same everywhere. Roles built around highly structured, repetitive tasks are more exposed when AI is used purely as a substitute. The picture changes when AI is used to augment human capability. In those cases, work isn’t simply reduced or streamlined; it becomes more deliberate. People and organizations are pushed to make conscious choices about what they do themselves and what they delegate to machines.

This shift demands a deep reskilling effort across the workforce, in two clear directions. First, employees need to learn how to work effectively with AI. That goes far beyond writing prompts. It means deciding which tasks to hand over and which to retain, validating outputs, building assistants that are genuinely useful, and integrating AI into workflows without losing control or quality. At the same time, the human capabilities that matter most in this context need to be strengthened. Clear judgment, effective communication, empathy, the ability to coordinate and prioritize work, negotiate and align interests, and a strong sense of purpose all become more important, especially when deciding not just how to work faster, but what is truly worth doing.

This shift demands a deep reskilling effort across the workforce, in two clear directions. First, employees need to learn how to work effectively with AI. That goes far beyond writing prompts. It means deciding which tasks to hand over and which to retain, validating outputs, building assistants that are genuinely useful, and integrating AI into workflows without losing control or quality. At the same time, the human capabilities that matter most in this context need to be strengthened. Clear judgment, effective communication, empathy, the ability to coordinate and prioritize work, negotiate and align interests, and a strong sense of purpose all become more important, especially when deciding not just how to work faster, but what is truly worth doing.

AI is no longer a technological challenge, but an organizational one. Companies that understand this will be able to capture its true value. Those that don’t will discover—too late—that automating without redesigning work isn’t transformation; it’s efficiency misunderstood.