IT Essentials: The customer is always right, unless you’re selling AI

Whose responsibility is it to make AI useful?

At Davos, Microsoft's Satya Nadella warned AI’s boom risks collapse unless it delivers real, widely shared value. Is it the responsibility of the “global community” to make this happen?

World leaders gathered at Davos last week for the World Economic Forum. Leavening the heavy toll of speeches marking the official collapse of the international rules-based order were various conversations with tech leaders. Guess what they were all about?

Amazon CEO Andy Jassy acknowledged that the multi-billion-dollar circular funding deals powering the build-out of AI infrastructure could prove risky longer term and admitted that Trump’s tariffs were starting to have an impact on some Amazon sellers.

Anthropic CEO Dario Amodei also had a dig at the Trump administration, stating that the decision to open exports of the Nvidia H200 chips to China was equivalent to selling nukes to North Korea but “bragging that Boeing made the casings.”

Echoing words we heard exactly one year ago about 2025, Open AI CFO Sarah Friar said that 2026 had to be the year of “practical adoption” for AI. That theme was developed by Microsoft CEO Satya Nadella, who told delegates that unless AIs benefits were more widely distributed, the boom would likely falter.

“We as a global community have to get to a point where we are using [AI] to do something useful that changes the outcomes of people and communities and countries and industries,” he said.

Well, yes. The problem is that, traditionally, if a product or service isn’t good or useful enough to make people and companies to want to buy it, we tend not to blame the customer. At least we didn’t used to. Not one but two pieces of research landed last week, both leading with the findings that greater use of AI tools in the workplace isn’t translating into healthier enterprise profit. Whose fault is that?

When viewed from a certain perspective Nadella’s words could be perceived as quite threatening.

“Adopt AI, at scale, NOW even though it still makes quite a lot of mistakes and it has the potential, if badly handled, to cause irreparable damage to your reputation, because if you don’t then the all the dodgy financial underpinnings will collapse like a house of cards, taking your pension pot with it and THEN YOU’LL BE SORRY.”

Nadella even addressed the environmental cost of datacentres. “We will quickly lose even the social permission to actually take something like energy, which is a scarce resource,” he said. This came as news to most of us, given that the likes of Nadella don’t have a track record of seeking permission for anything. Seeking forgiveness is more their MO.

Nadella is right to insist that AI must deliver real value and that its benefits be broadly shared if the social, economic, and environmental costs we will all bear for it are to be justified. Where we diverge is over who should shoulder the responsibility for ensuring this outcome.

It isn’t the responsibility of the “global community” to make AI useful. It’s the responsibility of the companies selling it.