Brad Smith (Microsoft): 'Ai is the new electricity but beware of the North-South divide'

Today, three years into the era of generative AI, Microsoft's data show that the pattern is repeating itself: 'a gap has rapidly emerged between the global North and South'. Currently, the use of Generative AI stands at 25 per cent in the global North and only 14 per cent in the South. The challenge, therefore, is not just to build AI, but to spread it, otherwise this technology will end up 'exacerbating and prolonging the gap that already exists'.

Huge physical resources are needed to run these algorithms. We are witnessing an "unprecedented investment in the technology sector in infrastructure", which translates into "data centres, GPUs, more broadband connectivity and electricity".

But Microsoft knows that lowering these silicon cathedrals from above creates friction. That is why Smith reiterates the importance of their 'Community First AI Infrastructure' initiative. When a big company comes to an area, communities have 'the reasonable right to ask tough questions'.

Redmond's response is based on precise commitments in five critical areas: 'electricity, water, jobs, taxes and skills'. A concrete example? Commitment to finance grid improvements so that consumers "don't see data centres increase the cost of electricity on the bills they pay".

On the European front, the burning issue remains data sovereignty. Prompted by fears that the US government could access data stored in Europe, Smith draws a clear line. He declares himself 'very confident that we protect the privacy' of European citizens.

How? Through technical solutions such as 'our sovereign controls', 'confidential computing' and data stored in the European Union. Microsoft includes 'strong safeguards in our contracts' and is ready to 'go to court or challenge national governments' if necessary.

In essence, the company acts as a legal and technological shield between European data and foreign government interference.

There is a key step that defines the maturity of a tech company: knowing when to stop. On the thorny topic of the use of AI in the military, Smith admits that there are huge complexities and that the industry cannot just push its products everywhere. On Wednesday, Microsoft officially sided with Anthropic by asking the Pentagon to suspend the blacklisting of the company founded by Dario Amodei , filing a brief in a San Francisco federal court in support of the Ai company's lawsuit. The blacklisting, it says, could have serious repercussions for many companies that supply technology to the US government. Microsoft

"The military," Brad Smith explained a few days before this stance was taken, "is one of the areas that raises the most complexities and questions about how Ai should be used. It is critical to ask not only where and how Ai should be used, but more importantly where it should not be used. Finding this balance is an essential issue, especially when it comes to military uses'

The crucial question, according to Microsoft's president, is not only how to implement the technology, but also its negation: "Where should Ai be used? How should AI be used? Where should it not be used?"

In response, Microsoft relies on published principles that create 'guard rails' (safety boundaries) within which the company is committed to remain. And if there is one thing that, according to Smith, must distinguish Microsoft from other Big Techs, it is the willingness to keep 'its feet on the ground'.

Finally, a note of reassurance for the markets and developers: despite Sam Altman's corporate upheavals, the recent conversion of OpenAI into a 'commercial company' does not change anything in the agreements made.