Peter Hinssen's The New Normal: what a 2010 book says about AI in 2026
In 2010 the Belgian technology thinker Peter Hinssen wrote that we were halfway through the digital revolution. The first half was about getting technology into people's hands. The second half would be about what happens once digital is simply normal. Sixteen years later we are deep into that second half, and AI is making the same move in a few years. The rules Hinssen wrote down then apply there too.
What Hinssen wrote in 2010
At the heart of The New Normal is a timeline. According to Hinssen, the first 25 years or so of the digital age were about adoption: computers, the internet and mobile phones reached consumers. In the 25 years after that, technology would become so self-evident that nobody, young or old, thinks about it anymore. That is what he calls the new normal.
A few rules follow from that. The most important: the more we depend on digital, the less we can afford it to fail. Customers have no patience left for digital failure, just as they have none for a power cut. Nor do they accept limits on price, speed, convenience or privacy, because they see elsewhere that it can be done differently.
For organizations, Hinssen argues, this means rethinking their information strategy, their technology strategy, the way they innovate and the way they are organized. Not as an IT project, but because customers and employees already live in the new normal and the business does not yet.
What came true
For everyday digital services that picture came true. Paying, banking, scheduling, communicating: most of it happens digitally, and nobody experiences that as innovation anymore. The rule about failure came true just as hard. An outage at a payment service or a cloud provider makes the news, and an SME that cannot invoice for a week because of ransomware feels it straight away in revenue. Digital is normal, so digital going down is a business problem, not an IT problem.
AI is taking the same path, only faster
The book is not about generative AI; it did not exist in this form in 2010. Still, it helps to hold AI against the same line. Where the internet and mobile took decades, AI took a few years. In 2023 an AI chat was an experiment. In 2026 AI comes built into Microsoft 365, the CRM and the accounting software, and employees have long used an AI assistant at home.
So Hinssen's rules apply to AI as well, and sooner than many businesses think:
- Nobody sees it as an experiment anymore. An employee who uses AI expects that it is allowed and that it works. Without agreements, it is in use anyway, just unseen.
- Tolerance for errors drops to zero. A wrong AI answer to a customer, a confidential document in a free tool, or an AI feature that is down for a day: none of that is seen as teething trouble anymore.
- Dependence shifts to suppliers. AI sits in other people's software. What they do with your data, and what happens when they stop or change their terms, is not up to you.
What this asks of an SME
Hinssen's point was that the organization has to catch up with what customers and employees already do. For AI that comes down to a few concrete things that do not need a big programme:
- Know which AI tools are used in your business, including the ones nobody approved, and record them in an AI register.
- Agree with employees which data may go into which tool, who checks the output, and what someone does when it goes wrong.
- Check per supplier what happens to your data and what your plan is if that supplier drops out.
- Know what you do when systems go down: which processes come back first, and whether your backups can actually be restored.
- For new uses involving personal data, check whether a DPIA is needed before it goes live.
None of these points is about the technology itself. They are about what Hinssen already pointed out in 2010: once something becomes normal, it becomes a basic utility, and a basic utility has to be in order.
The New Normal: Explore the Limits of the Digital World by Peter Hinssen was published in 2010. The summary above follows the book's description and reviews of it; applying it to AI is our own reading.