The CPF 2027 finance bill suggests a shift in who pays for skills training, placing more responsibility on employers, especially in areas related to critical thinking and artificial intelligence. As reported by Les Échos on September 23, employees would be expected to cover a larger portion of their training costs, increasing from 150 to 350 euros for most individuals. However, this increase would apply only after a review of the consistency of their professional development plans. Employers would still be responsible for funding training that is deemed a priority or co-financed by them. At this stage, no final decisions have been made on the proposal. According to the 2026 soft skills barometer by Lefebvre Dalloz Compétences, which surveyed 553 professionals, about 44% of companies have provided training in soft skills over the past year. These skills—such as communication, problem-solving, and adaptability—are increasingly seen as important, with companies investing more in evaluating them regularly. However, managers, while aware of the impact of AI on the workplace, often fail to fully recognize the importance of soft skills in successfully navigating this technological shift. As a result, more training is happening, but managers may not always know how to effectively assess these skills. One employee described an instance where a colleague used an AI tool to generate a report and then copied the entire text into their own document. A small note from the AI tool remained at the end, asking if the text was suitable or suggesting revisions. The document was then shared with this note still visible at the bottom. However, this detail was not captured during the annual performance review. Standard evaluation criteria focus on goals achieved, the quality of the final product, and teamwork. They do not examine the process between receiving the AI's response and submitting the work—such as proofreading, verifying accuracy, or expressing hesitation. This employee might receive a positive evaluation, and the manager would never be aware if critical thinking had been applied. Critical thinking is often seen as an abstract concept when described in general terms, but it becomes more tangible when linked to specific actions. Four work situations that reveal a step back from relying solely on AI include: asking for sources from the beginning, proofreading before sending, expressing doubt instead of automatically accepting the AI's output, and rephrasing instructions when the AI provides an incorrect response. These behaviors can be incorporated into performance interviews by asking questions such as: How did you verify this document? What sources did you ask the tool for? What made you doubt the result, and who did you consult? What changes did you make to your instructions when the answer didn’t fit? These questions give managers concrete insights and provide employees with the language to describe their own work. They also support the continuous evaluation efforts of companies, as these behaviors can be observed with each project, rather than waiting for an annual review.