In the current business ecosystem, Artificial Intelligence (AI) is no longer a futuristic promise to become the engine of a “great acceleration” that is already overcoming the human capacity to measure it. The growth is dizzying: ChatGPT has already reached 800 million weekly users (data from October 2025) having doubled – again – its base in just eight months. At the same time, 99% of companies in the Fortune 500 have already integrated AI technology into their processes. However, in the midst of this turmoil, a critical warning emerges: handing over strategic decisions to probabilistic algorithms is a mistake that could compromise the long-term survival of companies.
This is what the new CEMS Global Alliance report says – Augmented Leadership: Navigating the New Age of Intelligence (2025). According to the organization – which brings together 33 of the best universities and Management schools in the world, and of which Nova SBE is a member and the only representative from Portugal – this “acceleration” is not just technological: it is psychological pressure.
This is because the gap between adoption and real value creation is stark. A study by the Boston Consulting Group (BCG), cited by CEMS, reveals that only 5% of companies are generating real value with AI at scale, achieving revenue increases up to 5 times greater than their peers. In contrast, 60% of organizations report little or no impact despite massive investments. The error? Treating technology as an automatic decision maker rather than an amplifier of human intelligence.
AI is not rigorous, but we treat it as if it were
The root of the problem lies in the misunderstanding of the tool by those who use it. The potential of AI, especially generative Artificial Intelligence, is enormous, but its limitations are also immense and not understanding it can be fatal.
“A ChatGPT-style tool is not rigorous by definition. It works based on predictions and probabilities”, says Lénia Mestrinho, executive director of the Digital Data Design Institute at Nova SBE and Nova Medical School, in an interview with DN. “She can hallucinate with complete confidence”, recalls the teacher.
If a leader delegates a critical decision to a model that prioritizes ease of expression over precision, he or she is giving up truth in favor of probability. Lénia Mestrinho reinforces that literacy is the only antidote: “If we don’t have minimum literacy, it’s a very big risk.”
The error lies, therefore, in leadership that treats the “copilot” as if he were the infallible “commander”.
In technical terms, this process is called cognitive offloading. Inquiries from Forbes indicate that although 97% of business owners believe ChatGPT or a similar tool will help their companies, excessive dependence stunts critical thinking. The CEMS report warns that the leader runs the risk of becoming a mere “TikTok user”, consuming information passively.
“How does AI help us think, instead of giving us a ready-made answer?”, asks Lénia Mestrinho. Failing to synthesize information yourself takes away the leader’s ability to detect nuances. A leader who just signs what the AI wrote becomes a statistics validator. The rule is non-negotiable: “Think First, Prompt Second”. The human brain must be the initial filter and the final judge.
The “specialization gap” and the challenge of requalification
The visible face of management error manifests itself in radical cuts. The case of the Swedish financial company Klarna is paradigmatic: it announced its intention to reduce its workforce from 5,000 to 2,000 employees, claiming that AI does the work of thousands. However, CEMS raises a systemic issue: the expertise gap.
“It can take 10 to 15 years to build a high level of expertise,” says the study. By automating the “training ground” of juniors, companies are “eating the seed” of their future intellectual capital. Without these foundational experiences, where will the experts of 2040 come from?
This will give rise to the “boomerang effect”: companies that dispense with human resources in favor of AI will later face declines in originality and be forced to rehire humans at higher prices to recover the nuance that machines cannot simulate.
The scale of change is colossal. Lénia Mestrinho cites data from the World Economic Forum that estimates that around 40% of current skills will be replaced by 2030. In Europe, it is estimated that 100 million people will need to transform their digital skills in the next decade.
Leave a Reply