Microsoft CEO Calls for Caution Among Businesses on AI: They Risk Paying Twice

Microsoft CEO Satya Nadella has expressed a surprising stance on the future of artificial intelligence, warning companies that use proprietary AI models developed by large tech labs. According to Nadella, the main risk involves the possibility that enterprises unwittingly give up strategic information while leveraging these tools.

The executive argues that business users pay for artificial intelligence twice: first through the costs associated with using the models, and second through the sharing of internal data and knowledge necessary to achieve better results. The more a system is tailored to a company’s needs, the greater the amount of business information that enters the process.

Nadella highlights that models can learn from elements produced during everyday use, such as user requests, tools employed by automated agents, and corrections made by operators. These contributions, according to the CEO, can transform into highly valuable organizational knowledge that is difficult for a competitor to acquire.

Additional Details on Microsoft's (and Its CEO's) Stance Regarding AI

His stance also relates to the issue of the so-called "distillation" of models, that is, using the outputs of artificial intelligence to understand how the system works and develop alternative models. Nadella finds it contradictory that some companies can train their models using large amounts of public data while limiting others’ ability to analyze the results generated by their systems.

As a solution, Microsoft’s top executive proposes that businesses maintain control over their data, including messages, requests, and feedback provided to the models. He also suggests the creation of proprietary environments for learning and systems capable of handling multiple different AI models, avoiding excessive dependence on a single provider.

Meanwhile, interest in open-source models installed directly within corporate infrastructures is growing. According to some industry operators, many companies are considering these alternatives because they offer greater control over data and potentially lower costs compared to large commercial models. The trend seems to involve an increasing number of tech companies, with platforms dedicated to managing different AI models seeing a rise in the use of open solutions.