Microsoft 365 Copilot: Banning AI on Friday Afternoons?

Gartner's Warning on AI Usage at the End of the Week

The use of artificial intelligence tools, such as Microsoft 365 Copilot, could be banned on Friday afternoons. This was suggested by Dennis Xu, an analyst at Gartner, during a discussion on the security risks associated with these technologies. While his remarks focused on Microsoft's tool, his recommendations could extend to other AI platforms like ChatGPT, Gemini, or Claude.

The main reason for this warning lies in the fact that, despite the productivity gains offered by generative AI, the results require human verification. At the end of the week, employees may be more likely to skip this verification, thereby increasing the risk of errors.

The Persistent Errors of AI

Although artificial intelligence models continue to improve, they are not free from errors. It is therefore imperative to maintain systematic checks, particularly in the field of coding. A study published in March by Anthropic highlights the barriers to the widespread adoption of AI in certain professions, despite its theoretical capabilities to automate various tasks.

According to this study, some tasks that AI could theoretically perform are not being implemented due to the current limitations of the models. Other tasks take time to disseminate due to legal constraints, specific software needs, or the necessity for human verification.

This caution suggested by Dennis Xu and the observations from Anthropic underscore the importance of vigilance in the use of AI, especially at the end of the week when result verification may be overlooked.