AI makes decisions faster. Who will make them wiser?

The need for speed is well-understood in the business world. Speedy time-to-market, rapid cycle times, streamlined reporting chains, optimised dashboards and now with AI-led automation of some workflows, businesses may be able to make even faster decisions—at warp speed.

Indeed, some business leaders did make some rapid decisions and pronouncements.

In 2023, Bill Gates wrote that the risks of AI were “real but manageable”. In Davos the same year, Microsoft CEO Satya Nadella spoke of AI’s golden age and how it was good for humanity.

In 2024, Google rolled out its experimental AI Overviews, which told users to use glue to stick cheese on pizza and that geologists recommended one small rock be eaten every day for minerals. In the same year, the CEO of BNPL company Klarna spoke of how AI could do most jobs humans could do with their service chatbot doing the work of 700 employees. The workforce was reduced by 2,000, and hiring was paused for a year.

By mid-2025, Klarna was hiring humans back, as customer satisfaction plummeted and service quality was inconsistently delivered.

As I write in the summer of 2026, Bill Gates has warned of turbulence caused by AI, and that it will “either be the greatest equaliser ever invented, or the worst source of injustice”.

Also in 2026, Satya Nadella warned firms of how AI could threaten the very existence of a firm by taking their data, knowledge, thought process, ideas, and vision. Meanwhile, reportedly up to 70 per cent of Microsoft’s AI revenue comes from OpenAI.

The pronouncements and decisions about AI came fast. What has followed puts a question mark on the wisdom of those fast decisions without apparent consideration of potential risks—of second- or third-order effects of a new technology, or of such huge reliance on one company for revenue.

One could argue hindsight is 20/20. Stewardship, however, is a job of strategic foresight, such that it cannot afford fast and unwise decisions with undesirable impacts.

Gartner posits that by 2028, at least a third of enterprise decisions will be made autonomously or semi-autonomously by AI agents and 40 per cent of services will be AI-augmented.

These decisions could extend from real-time dynamic pricing to predictive capital allocation, fixing supply chain disruptions and solving customer service issues. These decisions are definitive and certain, and not always explainable.

In practice, however, a range of challenges plague enterprise automation journeys using generative AI, such as fragmented data, skills gaps, resistance to change in workflows, unclear ROI and governance gaps. These challenges are magnified in the Global South where the economies are complex and not always formalised, but rapidly growing; where inclusion is not a boardroom buzzword but a societal need for development; where the overall context is nuanced and often uncodified; and where the cost of mistakes is often human and devastating.

In one instance in the Global North, almost a year after warning about the over-reliance of the UK’s critical national infrastructure on three American providers of cloud, data, and models, Bank of England will now oversee Critical Third Parties.

Reality, however, can sometimes be harsher than our worst nightmares and imagination.

In August 2026, the world looked on in horror at the glacial collapse in Tibet and Nepal. The flood wave of water, pulverised ice, and rock that resulted reportedly “descended the first 22 kilometres (13.7 miles) at an average speed of 193 kilometres per hour (120 mph),” as per Jeffrey Kargel, a senior scientist at the Planetary Science Institute, quoted by CNN.

Some experts said it was not possible to predict it while others said if properly monitored, glaciers do show signs before they are about to crumble.

Something else happened too. As the gushing water and debris swept away bridges and houses, it lapped around a small, mint green house. Images and videos emerging from Nuwakot in Nepal’s Bagmati Province surprised everyone as the house stood steadfast with its residents on the balcony despite the provocation of the volume of water.

An engineering explanation was quickly found for its resilience. The use of reinforced concrete with steel rebars is said to have helped it absorb the lateral force as well as transfer some of the momentum to the foundation. Its location may also have helped it survive the onslaught of the mud.

In a region where houses are built with traditional stone masonry and non-reinforced concrete, somebody at some point had the foresight, the imagination, the risk sensibility and the engineering chops to make an unusual decision—about building this mint green house differently. It has proven to be a wise choice.

A group of researchers, primarily at UC San Diego, attempted to understand wisdom as a scientific construct, by drawing on empirical research data starting in the 1970s.

They defined wisdom as “a complex human characteristic or trait with specific components: social decision making, emotional regulation, prosocial behaviour, self-reflection, acceptance of uncertainty, decisiveness, and spirituality”.

A book titled Unsafe at Any Speed is older than that empirical data. Written about the American automobile industry that resisted the introduction of any safety features in cars, it was a non-fiction bestseller in 1966.

As speed is introduced into the automation of critical decisions in boardrooms and executive suites, it would be wise to ponder the core ideas in the book about the industry hype cycle, lobbying, other pressuring techniques, and rejection of technical criticism leaving customers unsafe.

Machine automation and AI may make us faster, but wisdom and the courage to make independent decisions for the long term will be essential to ensuring decisions remain safe at any speed. This is the kind of leadership now required to navigate the 'Uncharted Spaces' we are in.

The author is an experienced independent board director and the writer of 'Uncharted Spaces. Reset the Agenda. Reimagine the Boardroom.'.