Artificial intelligence can now produce research, analyses, and reports in seconds. But who verifies the reliability of its conclusions? The international firm PwC will launch a certification in critical thinking and professional judgment aimed at more than 85,000 employees and partners. Behind this announcement lies a central question: how to train tomorrow’s experts when AI handles part of the tasks that used to be learned?
The essentials
- Company : PwC US, whose program covers more than 85,000 partners and employees in the United States, Mexico, and the international service centers (Acceleration Centers).
- Timeline : start announced for December 2026; deployment and completion targeted by June 2027.
- Training : diagnostic of reasoning methods, masterclasses, customized simulations, and tools usable at work, with AI-supported guidance.
- Goal : learn to verify information, challenge hypotheses, and make evidence-based decisions.
Announced on November 9, 2026, the program named Critical Thinking & Judgment Certificate addresses a transformation of work that PwC regards as decisive. In its official communication, the group explains that professionals have long developed their judgment by seeking information, analyzing documents, testing hypotheses, and learning from their mistakes.
But these basic tasks are increasingly assisted or performed by AI tools. PwC believes it becomes necessary to deliberately create opportunities to exercise reasoning, rather than assuming that this skill automatically accrues with years of experience.
A large-scale program, launched in December 2026
The certification will cover more than 85,000 professionals within PwC US, including staff and partners working in offices in the United States and Mexico, as well as in its international service centers, called Acceleration Centers. It does not automatically cover all PwC global staff, a network that claims more than 360,000 people in more than 130 countries and territories.
The launch is planned for December 2026 and PwC aims for completion by June 2027. This timetable is an announced objective: the training is therefore not yet finished and its impact on the quality of decisions remains to be measured.
How to train an employee not to blindly trust AI?
The program will begin with a diagnostic of each participant’s reasoning habits. The objective is to highlight strengths, verification reflexes, and situations in which a professional might rush to a conclusion.
Participants will then undertake masterclasses and practical exercises, followed by simulations tailored to their profession. The training framework was developed by PwC with contributions from academic experts, including professors affiliated with the Wharton School and Harvard University. It is not, at this stage, about certifications issued by these two institutions.
In an example presented to Bloomberg Tax, an audit professional must examine an internal controls analysis produced by AI, identify any gaps, and challenge conclusions that are inadequately supported. The aim is not only to know how to use the tool, but to detect what it may have omitted, misinterpreted, or presented with excessive confidence.
An AI-based coach… to learn to manage AI
The program also includes an AI-based coach, trained in PwC’s pedagogical framework. According to details gathered by Human Resources Director, this tool will help employees prepare for difficult conversations, anticipate a client’s questions, or apply verification reflexes to their daily files.
This arrangement illustrates a paradox: the company uses AI technology to strengthen the human capabilities needed to govern it. The issue is not limited to the accuracy of a single answer, but concerns the ability to understand sources, recognize the limits of an analysis, and own a recommendation.
Not a traditional pass/fail exam
The certification would not be based solely on a final pass/fail exam. According to PwC leadership’s explanations reported by Business Insider, mastery of the skills should also be assessed through their application in daily work and professional evaluations.
The training must be tailored to the participants’ level of experience. A newly hired employee may learn to verify the sources of an analysis; an experienced partner may have to examine and challenge automated outputs produced for clients.
More than $100 million invested, according to Bloomberg
Asked by Bloomberg Tax, PwC estimates the total investment related to this initiative at more than $100 million, including development costs and funding for universities. This overall envelope does not correspond to a training expense of the same amount distributed exclusively among the 85,000 participants.
This choice underscores the importance the group places on human skills that become more visible as repetitive tasks and initial analyses are automated.
Universities also involved starting in 2027
PwC announces that from the beginning of 2027, its educational framework can be made available to any interested university. Within the framework of the Student Human Skills Initiative, a selection of institutions will also benefit from funding to develop teaching materials that incorporate these methods from introductory courses.
The stated objective is for students to learn to reason independently, evaluate contradictory information, and justify their decisions, including when using digital tools.
What lessons for Tunisian businesses and universities?
For banks, audit firms, IT services companies, and Tunisian educational institutions, this initiative raises a concrete question: if AI prepares an analysis, how do we still learn to control it? One possible answer is to integrate into training exercises where the student or employee must identify a missing source, compare several hypotheses, and explain the limits of a generated result.
This is a line of thought, not an announcement of PwC’s program adoption in Tunisia. The firm did not indicate, in its November 9 communication, that Tunisian companies or universities were already participating in this certification.
The challenge thus goes beyond simply learning prompts. As software produces documents and recommendations faster, organizations must ensure that their teams retain the time, skills, and responsibility needed to verify them.
