When Outlook goes down, we discover how dependent we really are

Yesterday, something remarkably simple became remarkably difficult for organisations around the world: sending an email.

A major Microsoft 365 outage that began on 12 November affected Outlook, Teams, Copilot and other connected services across multiple regions. At the peak, almost 130000 Outlook problems were reported through Downdetector, alongside tens of thousands of Microsoft 365 reports. Microsoft responded by redirecting traffic, restarting infrastructure and rolling back changes as it worked to restore services. 

The disruption stretched into a second day. Governments, businesses and healthcare organisations experienced problems accessing email and other services. Even integrations from AI providers such as OpenAI and Anthropic were affected because they depended on Microsoft infrastructure. 

There is an important detail here. Microsoft said it had no evidence that the incident resulted from malicious activity. Reports based on Microsoft’s incident information pointed instead to infrastructure capacity problems and an overloaded process responsible for establishing secure connections, while engineers also rolled back a recent network configuration change. 

That distinction matters. Because sometimes we learn more about cybersecurity and resilience from something that is not a cyberattack. The outage exposed something that normally remains almost invisible: the extraordinary concentration of our digital working lives.

Think about a typical working day. Email runs through Outlook. Meetings happen in Teams. Documents live in SharePoint or OneDrive. An increasing number of employees interact with Copilot. Identity, collaboration, communication and increasingly AI can all exist within the same ecosystem. When everything works, this integration is enormously convenient.

When part of that ecosystem fails, however, we suddenly discover all the invisible dependencies we have created. That is the bigger lesson from this outage.

For years, digital transformation has largely been presented as a story about efficiency. Move infrastructure to the cloud. Connect systems. Automate processes. Integrate communication. Add AI. Each step makes sense individually. Together, however, they can create concentration risk. An organisation may have hundreds of applications, thousands of devices and multiple communication channels while still depending on a surprisingly small number of underlying technology providers. Diversity at the interface does not necessarily mean diversity underneath.

AI makes this question even more interesting. We are currently racing to put AI assistants into almost every professional workflow. They can search our documents, summarise meetings, process email, prepare presentations and answer questions using organisational knowledge. But those capabilities do not exist in isolation. An AI assistant depends on identity services, cloud infrastructure, APIs, enterprise data, communication platforms and access controls. When one important layer disappears, capabilities elsewhere can disappear with it. During this outage, problems with Microsoft infrastructure were even felt in integrations offered by OpenAI and Anthropic.

The lesson, therefore, is not that we should stop using Microsoft 365, move everything out of the cloud or avoid AI. That would be the wrong conclusion. The lesson is that digital maturity cannot be measured only by how much technology we have adopted. It must also be measured by how well we can function when that technology is temporarily unavailable.

Cybersecurity professionals have used the concept of resilience for years. We create backups because storage can fail. We introduce redundancy because infrastructure can fail. We prepare incident-response plans because security controls can fail. Perhaps we need to apply the same thinking more broadly to the digital workplace.

If email disappears tomorrow morning, what happens? Can colleagues still communicate? Can students contact their lecturers? Can customers reach the organisation? Do employees know where they should look for status information if the system normally used to distribute that information is itself unavailable? And perhaps most importantly: have we ever actually tested any of this?

Those questions are particularly relevant for education.

Universities and colleges are becoming deeply digital organisations. Learning platforms, authentication, communication, documents, assessment, AI assistants and administrative processes increasingly depend on interconnected cloud services.

We teach students and employees how to use these systems, but digital literacy should perhaps include something else: understanding dependency.

Future professionals need to recognise that digital infrastructure is infrastructure. It can fail. Providers can experience outages. Connections can disappear. Automated processes can stop. AI assistants can suddenly become unavailable. Knowing what to do next is also a digital skill.

There is a parallel here with cybersecurity awareness. We have spent years teaching people what to do when they receive a suspicious email. Perhaps resilience education should also teach people what to do when there is no email at all.

None of this makes cloud computing, Microsoft 365 or AI a bad idea. The productivity and collaboration benefits are enormous. But mature technology adoption requires us to acknowledge the trade-off between integration and dependency. The more tightly connected our digital environments become, the more deliberately we must think about continuity.

Yesterday’s outage will eventually become another entry in a service-history log. Systems will recover, inboxes will fill again, Teams meetings will resume and most people will quickly forget about it. Organisations should not. Because the most valuable question after an outage is not simply:

“When will the service be back?”

It is:

“What did we discover we can no longer do without it?”

That is where the real conversation begins.

When the fridges start fighting back: what Europe’s supermarket cyberattacks teach us.

Cybersecurity is often discussed in abstract language. We speak about ransomware, zero-day vulnerabilities, attack surfaces and critical infrastructure. For many people, these concepts remain technical and distant until a cyberattack produces a visible consequence in everyday life.

Europe’s recent supermarket refrigeration incidents provide exactly such a moment.

On 2 November, attackers reportedly manipulated refrigeration systems at more than 100 stores across several European countries. Refrigerators and freezers were switched off, placed into defrost cycles or given dangerously high temperature settings. Retailers in Belgium, Luxembourg, Sweden, Ireland, Estonia and the Netherlands were among those affected. Some Swedish refrigeration systems were reportedly set to 50°C, while a Belgian butcher lost approximately €15000 worth of meat after refrigerator temperatures rose to 18°C.

The incidents affected stores connected to several major retailers, including Ahold Delhaize, Carrefour, Colruyt, ICA, Coop, SuperValu and even an IKEA location in Mons. Employees had to remove refrigerated products, move stock into refrigerated vehicles and discard food that could no longer be sold safely.

This was not an isolated event. In October, attackers reportedly interfered with refrigeration systems at at least 350 French supermarkets, including stores operated by Carrefour, Monoprix, Intermarché and Coopérative U. The affected systems included software from Danfoss, Copeland and Dixell.

At the time of writing, the perpetrators and their motives remain unclear. No credible group has publicly claimed responsibility, and the incidents should therefore not be attributed to a criminal organisation, state actor or hacktivist movement without further evidence.

Yet the identity of the attacker is not the most important lesson.

The more important question is why refrigeration equipment in ordinary stores could be reached and manipulated remotely in the first place.

Modern refrigeration installations are no longer isolated mechanical systems. They may contain connected sensors, controllers, monitoring platforms and remote-maintenance functions. These capabilities allow technicians to check temperatures, change settings and diagnose faults without travelling to every store. That can make maintenance faster and less expensive, but it also turns refrigeration into information technology.

Once a physical system is connected to a network, it inherits the risks of that network.

Researchers reportedly identified nearly 13000 publicly exposed refrigeration-management services across 99 countries and territories. Weak authentication, shared maintenance credentials, outdated equipment and direct exposure to the internet can leave these systems vulnerable. During the earlier French incidents, at least 4500 Danfoss interfaces were reportedly visible online. Some affected equipment had reached the end of its supported life or was still running outdated firmware.

This is where the incident becomes more than an unusual story about hacked refrigerators. It illustrates the growing convergence between information technology and operational technology.

Traditional IT manages information. Operational technology controls physical processes such as refrigeration, manufacturing, heating, transport, energy and water treatment. When an email account is compromised, information may be exposed. When operational technology is compromised, temperatures change, production stops, equipment overheats or physical goods are destroyed.

The boundary between digital and physical security has therefore become increasingly difficult to maintain.

For retailers, protecting refrigeration systems can no longer be left exclusively to refrigeration contractors. Procurement teams must ask whether equipment receives security updates, how long the manufacturer will provide support and whether remote access can be disabled or properly secured. IT departments need visibility into connected operational systems, while external technicians should receive individual, time-limited access rather than relying on shared passwords.

Network segmentation is equally important. A refrigeration controller should not be freely accessible from the public internet, nor should compromising it provide an easy route into payment, office or customer-information systems. Remote access should require strong authentication, preferably multifactor authentication, and unusual changes such as a freezer being set to 50°C should trigger an immediate alert or simply be rejected as physically unreasonable.

However, technical controls alone will not solve the problem. Employees, store managers, contractors, procurement specialists and executives must understand that apparently ordinary equipment can create cybersecurity risk. A refrigerator with remote monitoring is not merely an appliance. It is an endpoint. A maintenance account is not merely a convenience. It is a potential access path. Unsupported firmware is not merely an administrative inconvenience. It can become an operational and food-safety risk.

This has important implications for cybersecurity education. We still teach cybersecurity too often as if it happens primarily on laptops, servers and smartphones. Students and employees need to understand cyber-physical systems, supply-chain dependencies, remote maintenance and the connection between software vulnerabilities and physical consequences. That’s exactly what we teach at Howest.

The European refrigeration incidents make the lesson unusually concrete. Cybersecurity is no longer only about protecting information. It is about protecting the systems that keep food cold, buildings operational, factories running and essential services available.

Perhaps that is why this story deserves more attention than its initially strange headline suggests. When attackers can change the temperature of a supermarket freezer from another location, the distinction between the digital and physical world has effectively disappeared. Our cybersecurity culture, education and investment must now catch up.

Stop teaching students to use AI. Teach them to think with AI.

Higher education has a habit of turning technological change into a curriculum problem. A new technology arrives, employers start talking about it, students begin experimenting with it, and sooner or later we ask the inevitable question: where should we teach this?

Artificial intelligence is following exactly that pattern. We introduce students to generative AI. We organise workshops about prompting. We demonstrate AI assistants and agents. We discuss which tools students may or may not use for assignments. All of these initiatives have value, but I increasingly wonder whether we are focusing on the wrong educational challenge.

Teaching students how to use AI is relatively easy. Teaching them how to think with AI is much harder.

The difference matters because using an AI tool requires surprisingly little expertise. Modern AI systems are deliberately designed to make sophisticated capabilities accessible through increasingly simple interfaces. Students can generate text, analyse documents, create software, brainstorm ideas and build basic AI-powered workflows with remarkably little technical knowledge.

That accessibility is one of AI’s great strengths, but from an educational perspective it creates a paradox. The easier AI becomes to operate, the less useful operating the technology becomes as evidence that somebody understands what they are doing.

Consider a student using an AI agent to automate a business process. The impressive part is not necessarily building the agent. Increasingly, AI can help the student do that too. The difficult questions come before the technology: What problem are we solving? How does the existing process work? Why does it work that way? Which parts should be improved? Which decisions should remain human? What could go wrong if we automate them?

Only after answering those questions should we ask what technology belongs underneath them.

This distinction became particularly clear to me during a recent discussion about developing an interdisciplinary student project around AI agents. The learning journey we discussed did not begin with the agent. It began with understanding a business process, documenting it and identifying opportunities to improve it. Automation came afterwards.

I think there is a much larger educational principle hidden inside that sequence.

Understand. Question. Improve. Then automate.

Too often, AI education starts at the end. With the technology. We give students a powerful tool and teach them what buttons to press, which prompts to enter or how to connect different systems. They may produce an impressive result, but that does not automatically mean meaningful learning has taken place.

The European debate around AI literacy is fortunately moving in a broader direction. The Council of Europe’s framework approaches AI not merely as a technology but as something embedded in institutions and society, explicitly connecting AI literacy with human rights, democracy, critical thinking and social agency. During UNESCO’s Global Media and Information Literacy Week, the emphasis is similarly on empowering people to navigate information critically and contribute to a trustworthy information environment.

That points towards a much richer definition of AI literacy. AI literacy is not knowing five clever prompts. It is understanding enough about AI to make informed decisions around it. It involves questioning outputs, recognising limitations, evaluating information and deciding when human judgment must take precedence. And this is where higher education has an extraordinary opportunity. If AI can increasingly perform parts of the work our graduates will encounter, universities should not respond simply by teaching students to perform those same tasks with AI. We should move further up the cognitive chain.

Instead of only asking students to produce an answer, ask them to evaluate one. Instead of only asking them to generate a solution, ask them to compare alternatives. Instead of rewarding the finished product alone, examine the decisions that produced it. Instead of prohibiting AI because it makes an assignment too easy, perhaps we should reconsider why the assignment became easy in the first place.

This changes the role of the educator as well. Our job cannot simply be transferring knowledge that students can retrieve within seconds. Increasingly, education must create environments where students learn to interpret, challenge, apply and defend knowledge. That includes knowing when not to use AI.

There is another reason why this matters. Today’s students will graduate into organisations that are themselves struggling with these questions. Employers do not simply need people who know how to access AI. Their employees already have access. What organisations increasingly need are people capable of using AI without outsourcing their judgment to it. That requires subject knowledge. It requires critical thinking. It requires ethical awareness. It requires curiosity and the confidence to challenge a convincing machine-generated answer.

Those abilities will outlive any particular AI platform.

Today’s students may use ChatGPT, Copilot, Claude or another system. Five years from now, the interfaces and underlying technologies may be completely different. Building an education strategy around today’s tools therefore has a very short shelf life. Building education around judgment does not. Perhaps this is the real challenge AI presents to education. It forces us to distinguish between completing a task and understanding a task, between generating an answer and knowing why it is a good answer.

So yes, we should teach students how to use AI. But that is the easy part. Our real responsibility is teaching them how to remain intellectually in control while using it. Because the future will not belong to graduates who simply know how to operate AI. It will belong to those who know what to ask, what to question, what to trust and what should remain human.

AI Literacy is no longer optional. The AI Act makes that clear

For years, discussions about artificial intelligence focused almost exclusively on technology. Organisations debated which tools to adopt, how much productivity they could gain, and whether AI would disrupt industries or replace jobs. The assumption was often that the biggest challenge would be technological implementation: selecting the right systems, integrating them into existing processes and staying ahead of competitors.

Today, that assumption increasingly looks incomplete.

As AI becomes embedded in everyday work, the critical question is no longer whether organisations have access to AI. Most already do. Employees are using generative AI to write reports, draft emails, analyse information, create presentations and support decision-making. In many cases, AI capabilities have arrived faster than organisations anticipated. What has not always kept pace is people’s understanding of how to use these systems responsibly.

This is precisely where the European AI Act introduces an important shift in perspective.

While much of the public discussion surrounding the AI Act focuses on risk classifications and compliance obligations, one of its most significant messages receives far less attention: people who work with AI need a sufficient level of AI literacy. In other words, responsibility does not end with deploying a system. Organisations must also ensure that the people interacting with that system understand its capabilities, limitations and risks.

This may ultimately prove to be one of the most consequential aspects of the legislation. The AI Act is effectively acknowledging something that many educators have argued for years: the success of AI adoption depends less on the technology itself than on the people using it.

That insight should sound familiar.

Two decades ago, cybersecurity faced a similar evolution. In the early days, security was largely considered a technical responsibility. Firewalls, antivirus software and network protection were expected to solve the problem. Employees were often viewed as passive users operating within a secure system.

Over time, organisations learned a difficult lesson. Many of the most serious security incidents were not caused by technical failures but by human behaviour. A single employee clicking a malicious link could bypass millions of euros worth of security investments. As a result, cybersecurity evolved from a purely technical discipline into an organisational responsibility. Awareness programmes, phishing simulations and security training became standard practice.

We rarely question that approach today. No serious organisation would expect employees to work safely in a digital environment without a basic understanding of cybersecurity. AI is now following the same path.

The risks associated with AI are often discussed as technical challenges, yet many of them are fundamentally human challenges. An employee accepts an AI-generated answer without verification. A manager relies on an AI-generated recommendation without understanding its limitations. Sensitive information is entered into an external AI system. An automatically generated summary is treated as fact despite containing critical inaccuracies.

In each of these cases, the technology may function exactly as designed. The problem lies elsewhere. The problem is that the user lacks the knowledge, judgment or awareness required to use the technology responsibly. This is why AI literacy cannot be reduced to learning how to write better prompts.

Prompting is useful, but it represents only a small part of the challenge. True AI literacy involves understanding how AI systems generate outputs, recognising when those outputs may be unreliable, evaluating the quality of generated information, understanding ethical implications and maintaining accountability for decisions. It requires people to understand not only what AI can do, but also where its limitations begin.

The distinction matters because organisations are increasingly approaching AI as a technology initiative when, in reality, it is also a workforce development initiative.

Every organisation that adopts AI is, whether intentionally or not, changing the competencies its employees require. The challenge is not limited to technical teams. Human resources professionals need to understand AI when evaluating recruitment tools. Marketing teams need to understand AI when using generated content. Managers need to understand AI when interpreting AI-assisted insights. Executives need to understand AI when making strategic decisions about governance and risk.

AI literacy is rapidly becoming a universal workplace competence rather than a specialist skill.

This has implications for organisational learning. The question is no longer whether organisations should provide AI training. Increasingly, the question is whether they can afford not to.

The organisations that thrive over the coming decade will likely not be those that simply deploy the most AI. Access to powerful AI systems will become increasingly common. Competitive advantage rarely comes from technology alone once that technology becomes widely available.

Instead, the differentiator will be people.

  • People who know how to question an AI-generated answer.
  • People who understand when verification is required.
  • People who recognise risks before they become incidents.
  • People who can combine human judgment with machine intelligence.

In that sense, the AI Act should not be viewed merely as a compliance exercise. It is an early signal of a much larger transformation. The regulation recognises that responsible AI adoption depends on capable people. Technology may enable the future, but literacy determines whether that future creates value or risk.

This is also why higher education cannot remain a spectator. If organisations are increasingly expected to develop AI-literate employees, then universities and colleges must help develop AI-literate graduates. Students should not encounter these competencies for the first time during corporate compliance training. They should enter the workforce already understanding how to work critically, responsibly and effectively with AI.

For years, digital literacy was considered a prerequisite for participation in modern society. We may now be reaching a similar moment with AI literacy.

The European AI Act is perhaps the clearest indication yet that AI literacy is no longer an educational aspiration or an optional professional development topic. It is becoming a foundational competence for the modern workplace. And the organisations that recognise this early will be far better prepared for the future than those that focus solely on the technology itself.

The future will not belong to organisations with the most AI. It will belong to organisations with the most AI-literate people.

For organisations, this creates both a challenge and an opportunity. The challenge is clear: AI literacy can no longer be treated as an optional skill reserved for technical specialists. The opportunity is that organisations that invest early in AI literacy will not only be better prepared for compliance requirements, but will also build a workforce capable of using AI more effectively, responsibly and strategically.

As organisations begin to translate the requirements of the AI Act into practical learning initiatives, many are discovering that AI literacy is not a one-off workshop or a checklist exercise. It requires a structured approach that combines awareness, critical thinking, ethics, governance and practical application. That is precisely why I spend much of my time helping organisations, educational institutions and public sector organisations develop AI literacy programmes tailored to their context. Through training, workshops and my books on AI literacy and responsible AI adoption, I help teams move beyond the hype and build the knowledge and judgement needed to work confidently with AI.

Because ultimately, the organisations that will benefit most from AI will not be those with access to the most advanced tools. They will be those whose people understand how to use those tools wisely.

Human Agency: The Most Valuable Skill in the Age of AI

A shift is underway in how we think about technology, education, and work. For decades, the primary objective was to help people learn how to use increasingly sophisticated technologies. Digital literacy became a critical competency, and success often depended on one’s ability to navigate new tools and systems. Today, however, the challenge is changing. The question is no longer simply whether people can use technology. Instead, it is whether they can remain distinctly human while working alongside increasingly capable artificial intelligence systems.

This distinction is more important than many people realize.

When generative AI entered the mainstream, much of the conversation focused on productivity. Organizations and individuals quickly began asking practical questions. Could AI write reports, produce presentations, summarize documents, generate software code, or assist with research? The answer, in many cases, turned out to be yes. Not only could AI perform these tasks, but it could often do so with remarkable speed and surprising quality.

Yet as AI capabilities advanced, something unexpected became clear. The primary limitation was no longer technological. The true bottleneck was human judgment.

Artificial intelligence can generate answers, but it cannot ultimately determine whether those answers are correct. It can produce recommendations, but humans remain responsible for deciding which actions should be taken. AI can craft persuasive messages, analyze large volumes of information, and even simulate expertise. However, it cannot replace the human responsibility to assess appropriateness, evaluate risks, or consider ethical consequences.

For a brief period, many believed that the rise of AI would primarily create a technical challenge. Increasingly, evidence suggests that the greater challenge is a human one. The future will belong not simply to those who know how to operate AI tools, but to those who understand when to trust them, when to question them, and when to disregard them altogether.

This is where the concept of human agency becomes essential.

Human agency refers to the capacity to act intentionally, make informed decisions, and accept responsibility for outcomes. It is the quality that enables individuals to remain active participants rather than passive consumers of automated outputs. In an AI-driven society, agency is not becoming obsolete. Instead, it is becoming one of the most valuable capabilities a person can possess.

The implications are particularly visible in education. Students today have access to powerful AI systems that can explain concepts, answer questions, generate essays, design presentations, and provide instant feedback. As a result, the educational challenge is evolving. The key question is no longer whether students can produce an answer. The more important question is whether they truly understand it.

Can they evaluate the quality of the response? Can they recognize flaws, inaccuracies, or bias? Can they improve upon it and defend their reasoning? Most importantly, can they identify when the answer is wrong?

These concerns extend far beyond the classroom. The same transformation is taking place in the workplace. Employees increasingly rely on AI to draft emails, summarize meetings, analyze data, and generate creative ideas. Once again, the challenge is no longer production. It is evaluation.

Individuals who accept AI-generated outputs without any doubt introduce new risks into organizations. By contrast, those who combine AI capabilities with critical thinking, contextual awareness, and ethical judgment create genuine value. In this environment, competitive advantage stems not from automation itself, but from the quality of human oversight applied to it.

This shift carries significant implications for education systems. For many years, educational institutions have focused primarily on the transmission of knowledge. While knowledge remains important, the growing presence of AI means that capability must receive equal attention. Students need opportunities to develop critical thinking skills, formulate better questions, assess the reliability of information, understand bias, and navigate uncertainty. Above all, they need the confidence and intellectual independence to challenge answers generated by machines.

In many ways, artificial intelligence is forcing society to rediscover the value of deeply human capabilities. Curiosity, judgment, creativity, reflection, communication, and ethics have long been described as “soft skills.” Yet that label increasingly feels outdated. In a world where machines can automate many routine tasks, these qualities are becoming strategic skills that directly influence organizational success and individual effectiveness.

History offers an important lesson in this regard. Whenever technological progress automates routine work, the value of uniquely human capabilities tends to increase. The Industrial Revolution elevated the importance of management and design. The digital revolution increased the value of innovation and problem-solving. The age of AI appears to be following a similar pattern. As machines become more capable of generating content and processing information, the importance of human judgment rises accordingly.

The professionals who thrive in the coming years will not be those who compete with AI on its own terms. They will be those who learn how to direct, guide, and challenge it effectively. Likewise, the educational institutions that succeed will not be those that attempt to prohibit AI. They will be those that equip students with the agency required to use it responsibly and intelligently. And the organizations that excel will not necessarily be those with the greatest number of AI tools. They will be those with people who are capable of making sound decisions in AI-augmented environments.

This is why we may be asking the wrong question. Too much attention continues to focus on what AI can do. Not enough attention is devoted to what humans should continue doing.

The future will not be shaped by artificial intelligence alone. It will be shaped by the quality of human judgment that guides it. Technology may provide unprecedented capabilities, but it is human agency that determines how those capabilities are used, which outcomes are pursued, and which values are preserved.

For that reason, human agency may well become one of the most important skills of the twenty-first century. In an age defined by increasingly intelligent machines, our greatest advantage may ultimately be our ability to think, decide, question, and act with purpose.

Waarom we moeten stoppen met studenten op te leiden tot programmeurs

De discussie over artificiële intelligentie en softwareontwikkeling verloopt opvallend voorspelbaar. Zodra iemand suggereert dat AI een fundamentele impact zal hebben op IT-jobs, volgt vrijwel onmiddellijk een geruststellende reactie vanuit de sector. Er zullen altijd ontwikkelaars nodig zijn. AI zal programmeurs ondersteunen, niet vervangen. Software wordt immers steeds belangrijker, dus ook de mensen die software bouwen zullen nodig blijven.

Hoewel daar ongetwijfeld een kern van waarheid in zit, heb ik steeds meer het gevoel dat we daardoor de essentie van de discussie missen. De vraag is immers niet of organisaties in de toekomst nog technische profielen nodig zullen hebben. De vraag is of de technische profielen die we vandaag opleiden dezelfde rollen zullen opnemen als de technische profielen van morgen.

Dat onderscheid lijkt subtiel, maar het maakt een wereld van verschil.

Decennialang werd de waarde van een IT-professional in belangrijke mate bepaald door zijn of haar vermogen om software te bouwen. Bedrijven hadden nood aan toepassingen, websites, databases, rapporteringstools en integraties tussen systemen, maar beschikten zelf niet over de kennis om die oplossingen te realiseren. Daardoor ontstond een duidelijke scheiding tussen business en IT. Aan de ene kant stonden de mensen die de problemen kenden, aan de andere kant de mensen die over de technische vaardigheden beschikten om die problemen op te lossen. De programmeur vormde de brug tussen beide werelden.

Dat model werkte goed omdat programmeren een schaars talent was. Voor de meeste mensen was softwareontwikkeling een complex vakgebied dat jaren training vereiste. Wie code kon schrijven, bezat kennis die voor de meeste collega’s ontoegankelijk was. De technische implementatie vormde daardoor vaak de grootste uitdaging binnen een project.

Vandaag zien we echter iets ontstaan wat we in de geschiedenis van softwareontwikkeling nog niet eerder hebben meegemaakt. Voor het eerst begint de technische vertaalslag zelf te automatiseren. Steeds krachtigere AI-systemen zijn in staat om software te genereren op basis van natuurlijke taal, bestaande code te analyseren, fouten op te sporen, documentatie te schrijven en zelfs volledige toepassingen op te bouwen vanuit een relatief beperkte beschrijving van een probleem. De technologie bevindt zich nog lang niet op een niveau waarop menselijke tussenkomst overbodig wordt, maar ze is ondertussen wel krachtig genoeg om de aard van softwareontwikkeling fundamenteel te veranderen.

Dat betekent niet dat programmeurs zullen verdwijnen. Het betekent wel dat programmeren steeds minder de activiteit wordt waarvoor organisaties bereid zijn te betalen.

Dat klinkt misschien provocerend, maar eigenlijk is het een logisch economisch principe. Naarmate iets goedkoper en toegankelijker wordt, verschuift de waarde naar een hoger niveau in de keten. Toen websites bouwen een zeldzame expertise was, lag de waarde bij de webdeveloper. Vandaag zijn er talloze platformen waarmee een website in enkele uren kan worden samengesteld. Daardoor ligt de meerwaarde niet langer in het bouwen van de website zelf, maar in het begrijpen van de doelgroep, het ontwerpen van een goede gebruikerservaring en het creëren van relevante inhoud.

Dezelfde beweging zien we nu ontstaan binnen softwareontwikkeling.

Veel organisaties kampen vandaag niet met een tekort aan code. Ze kampen met een tekort aan mensen die begrijpen waar de echte problemen zich bevinden. Ze worstelen met inefficiënte processen, versnipperde informatiestromen, onduidelijke verantwoordelijkheden en afdelingen die elkaar moeilijk begrijpen. Geen van die problemen wordt opgelost door snel extra code te produceren. Ze vereisen mensen die voldoende technisch inzicht hebben om oplossingen te ontwerpen, maar ook voldoende organisatorisch inzicht om de onderliggende uitdaging correct te analyseren.

Precies daar zie ik de grootste verschuiving voor toekomstige IT-profielen.

Wanneer een steeds groter deel van de technische implementatie wordt ondersteund door AI, verschuift de menselijke meerwaarde automatisch naar activiteiten die voorafgaan aan het bouwen van de oplossing. Het vermogen om processen te begrijpen wordt belangrijker. Het vermogen om de juiste vragen te stellen wordt belangrijker. Het vermogen om gebruikers te begeleiden, tegenstrijdige belangen te verzoenen en technologie af te stemmen op strategische doelstellingen wordt belangrijker.

Opvallend genoeg zijn dit vaardigheden die binnen veel technische opleidingen traditioneel minder aandacht krijgen dan de puur technische componenten. Studenten investeren honderden uren in programmeertalen, frameworks en ontwikkelmethodologieën, terwijl communicatie, procesanalyse, consultancyvaardigheden en organisatorisch inzicht vaak een meer bescheiden plaats krijgen binnen het curriculum. Dat was jarenlang verdedigbaar omdat de technische kennis zelf de grootste onderscheidende factor vormde. De vraag is echter of dat in een AI-gedreven context nog steeds het geval zal zijn.

Misschien moeten we beginnen erkennen dat veel van de competenties die jarenlang als “soft skills” werden beschouwd, in werkelijkheid strategische kernvaardigheden zijn. Een technisch profiel dat een complexe organisatie kan doorgronden, noden kan detecteren die nog niet expliciet geformuleerd zijn en verschillende stakeholders rond een gezamenlijk doel kan verzamelen, creëert immers een vorm van waarde die veel moeilijker te automatiseren is dan het schrijven van een functie of het bouwen van een gebruikersinterface.

Daarom vraag ik me steeds vaker af of het woord “programmeur” nog wel de juiste term is voor de professionals die we vandaag opleiden. Het woord suggereert dat het schrijven van code centraal staat. Het definieert een professional op basis van zijn gereedschap in plaats van op basis van zijn maatschappelijke of economische bijdrage. We noemen architecten ook geen CAD-operatoren en artsen geen invullers van medische dossiers. Hun waarde ligt niet in de instrumenten die ze gebruiken, maar in hun vermogen om complexe problemen op te lossen binnen hun domein.

Misschien moeten we technische profielen op dezelfde manier bekijken. De professionals die organisaties in de toekomst nodig hebben, zijn niet noodzakelijk degenen die de meeste code kunnen produceren. Organisaties hebben nood aan mensen die technologie begrijpen, maar tegelijk kunnen redeneren vanuit processen, gebruikers, bedrijfsdoelstellingen en maatschappelijke context. Ze hebben nood aan mensen die kunnen bepalen wanneer AI een oplossing is, wanneer klassieke software volstaat en wanneer het probleem helemaal niet technisch van aard is. Ze hebben nood aan professionals die kunnen functioneren op het raakvlak tussen technologie, business en menselijke interactie.

Dat betekent niet dat technische expertise minder belangrijk wordt. Integendeel, een grondig begrip van software, data, infrastructuur en AI blijft essentieel. Het betekent wel dat die expertise steeds vaker een middel wordt in plaats van een einddoel. De echte uitdaging verschuift van “hoe bouwen we dit?” naar “wat moeten we eigenlijk bouwen en waarom?”

Misschien zullen we binnen tien jaar nog altijd evenveel mensen nodig hebben in de technologiesector als vandaag. Misschien zullen het er zelfs meer zijn. Ik vermoed echter dat hun functieprofielen veel minder zullen draaien rond programmeren en veel meer rond het ontwerpen van oplossingen, het verbeteren van processen en het verbinden van mensen, technologie en organisatiedoelstellingen.

We hebben nood aan vertalers tussen menselijke behoeften en digitale mogelijkheden.

De grote fout die we momenteel dreigen te maken, is dat we blijven denken in termen van programmeertalen terwijl de markt steeds meer nood krijgt aan vertalers. Niet vertalers tussen Nederlands en Engels, maar vertalers tussen business en technologie, tussen menselijke behoeften en digitale mogelijkheden, tussen complexe organisaties en de systemen die hen moeten ondersteunen.

Dat zijn misschien niet de beroepen waarmee veel IT-studenten ooit aan hun opleiding begonnen. Het zijn wel de beroepen waarvan ik vermoed dat ze de komende jaren de grootste impact zullen hebben.

Want hoe beter AI wordt in programmeren, hoe duidelijker één waarheid naar voren komt: code was nooit het uiteindelijke product. Code was slechts de manier waarop we oplossingen bouwden. De echte waarde lag altijd in het begrijpen van het probleem. En precies daarom denk ik dat de opleidingen die vandaag nog programmeurs proberen af te leveren, het risico lopen de professionals van gisteren op te leiden, terwijl de organisaties van morgen op zoek zullen zijn naar iets heel anders.

Beyond Erasmus: Why international partnerships are becoming a competitive advantage for universities

A few days ago, we welcomed a delegation from six Danish universities to Howest. The visit lasted only a few days, but the conversations stayed with me long after our guests had returned home.

When people hear the word internationalisation, they often think about student exchanges, study abroad programmes or Erasmus mobility. Those remain valuable and they always will. But after spending time with colleagues from Denmark, I was reminded that internationalisation is becoming something much bigger.

The future of higher education is increasingly being built through international networks of institutions that learn from each other, innovate together and tackle shared challenges collectively. And that matters more than ever.


The challenges we face are no longer local

During the discussions with our Danish colleagues, I was struck by how similar our conversations were. Whether we were talking about AI in education, digital transformation, skills shortages, student engagement, international mobility or curriculum design, the same themes kept emerging.

How do we prepare students for jobs that are changing faster than ever? How should educators respond to generative AI? Which skills will remain valuable when technology can automate more tasks? How do we keep higher education relevant for society and business?

These are not Belgian questions. They are European questions. And increasingly, they are global questions. No university, no matter how strong, can solve them alone.


Why visits like this matter

It is easy to underestimate the value of hosting international visitors. From the outside, it can look like another official delegation, another series of presentations, another campus tour. But the real value lies in something less visible. The conversations in between. The exchange of experiences. The moments where someone describes a challenge they are facing and you realise:

“We’re struggling with exactly the same thing.”

Or the opposite:

“That’s an interesting approach. Why aren’t we doing that?”

Those moments accelerate learning. A good international partnership allows institutions to learn in days what might otherwise take years of trial and error.


Howest is not just participating. We are contributing.

One thing made me particularly proud during the Danish visit. The conversations were not one-directional. We were not simply listening and learning. We were also sharing.

Over the past years, Howest has built a strong reputation in areas such as applied IT education, cybersecurity, internationalisation, innovation projects and industry collaboration. More recently, we have been investing heavily in AI literacy and helping students develop the skills they need to work effectively, critically and responsibly with AI. These were topics that generated significant interest among our visitors. That is important.

Higher education is often described as a knowledge ecosystem. But ecosystems only work when knowledge flows in multiple directions.

Strong partnerships are not built on one institution teaching another. They are built on mutual exchange.


AI is making international collaboration more important

If there was one topic that appeared in almost every conversation during the visit, it was artificial intelligence. Not because everyone was talking about the latest AI tools. But because everyone is trying to answer the same educational questions.

  • How should we assess students?
  • How do we teach critical thinking when AI can generate plausible answers in seconds?
  • What does AI literacy actually mean?
  • How do we prepare graduates for workplaces where AI becomes part of everyday work?

The reality is simple. None of us have all the answers. And that is precisely why international cooperation matters. The pace of technological change is simply too fast for institutions to work in isolation. We need networks. We need partners. We need opportunities to share what works and what doesn’t.


Internationalisation is becoming a driver of innovation

For a long time, internationalisation was often viewed as a separate activity. Something managed by an international office. Something measured through mobility numbers. That view feels increasingly outdated. Today, internationalisation is becoming intertwined with innovation.

The best partnerships create opportunities for:

  • Joint research
  • Shared educational innovation
  • New learning experiences
  • Industry collaboration
  • Skills development
  • Talent mobility
  • European funding opportunities

In other words, internationalisation is becoming a strategic tool for institutional development. Not an add-on. A driver.


What students gain

Students may never see most of the meetings, discussions and partnerships that happen behind the scenes. But they experience the outcomes. They gain access to broader networks. New opportunities, international projects. Different perspectives. Connections with peers and professionals from across Europe.

In a world where careers are becoming increasingly international and digital, those experiences matter.

The graduates entering the labour market today will collaborate with people from different countries, cultures and disciplines throughout their careers. Universities have a responsibility to prepare them for that reality.


Looking ahead

The recent visit from the Danish universities left me optimistic. Not because it produced a spectacular announcement. Not because it generated a headline. But because it demonstrated something important. Across Europe, universities are wrestling with remarkably similar questions about AI, skills, innovation and the future of education. And instead of trying to solve them independently, more institutions are choosing to tackle them together.

That is good news.

Because the challenges ahead are becoming increasingly complex. But so are the networks we are building to address them. For institutions like Howest, internationalisation is no longer just about crossing borders. It is about creating connections that help students, staff and organisations learn faster, innovate better and prepare for a future that none of us can fully predict on our own. And after spending a few days exchanging ideas with our Danish colleagues, I am more convinced than ever that those connections will be among the most valuable assets a university can have.

AI Literacy, Digital Skills and Innovation: What Belgium should know this week

If you only have ten minutes this weekend to catch up on technology, education and innovation news, these are the developments I’d focus on.

What stood out this week is that many conversations are no longer about whether AI will be adopted. The discussion is increasingly about governance, skills, infrastructure and responsible deployment. That shift matters. It signals a move from experimentation to implementation, with consequences for schools, universities, businesses and policymakers.


1. The European Commission publishes new evidence on AI in education

What happened?

On 22 September, the European Commission published two major reports examining the impact of generative AI on learning and teaching, and the broader state of digital education across Europe. The reports conclude that generative AI is steadily moving into mainstream educational practice, but that evidence on long-term learning outcomes remains mixed and teacher training remains a major challenge. The Commission also found that nearly two-thirds of European education systems now have AI-related policy frameworks.
Source: European Commission
https://digital-strategy.ec.europa.eu/en/news/two-commission-reports-show-impact-artificial-intelligence-and-digital-technologies-teaching-and

Why does it matter?

This is one of the clearest signals yet that Europe’s AI conversation is moving beyond technology and into pedagogy. The Commission is essentially saying: AI is here, but we still do not fully understand its impact on critical thinking, learning habits and educational quality.

Why should Belgian readers care?

Belgian schools and universities are wrestling with exactly these questions. Many institutions have introduced AI guidelines during the past two years, but relatively few have fundamentally redesigned curricula or assessment. The Commission’s findings suggest that professional development for teachers may become one of the most important education investments of the next decade.


2. AI Literacy Is Becoming a Strategic Priority in Higher Education

What happened?

During the opening of the academic year, the Vrije Universiteit Brussel (VUB) emphasised that AI literacy should become a core competency for every graduate. The university argued that students need to understand not only how AI works, but also how to use it critically, responsibly and effectively.

Why does it matter?

The conversation in higher education is gradually shifting from “Can students use AI?” to “How do we prepare students to work with AI?”

That may sound like a subtle change, but it has major implications for curriculum design, assessment methods and graduate skills. It suggests that institutions need more than AI guidelines. They need learning outcomes, teaching approaches and evaluation strategies that reflect a world where AI is becoming an everyday tool in both professional and academic environments.

Perhaps the most important insight is that AI literacy is not just about technology. It combines technical understanding, critical thinking, ethical awareness, information literacy and the ability to judge when human expertise remains essential.

Why should Belgian readers care?

VUB is not alone in moving in this direction. Across Belgium, universities and universities of applied sciences are exploring how AI competencies should be embedded into education.

At Howest, for example, significant efforts are underway to develop an AI Literacy learning pathway for students. Rather than treating AI as a standalone technology topic, the initiative focuses on practical skills that students can immediately apply throughout their studies and future careers. These include working effectively with generative AI, prompt engineering, critically evaluating AI-generated content, understanding ethical and legal considerations, and recognising the impact of AI on professional practice.

What makes this particularly relevant is that employers increasingly expect graduates to possess these capabilities. Technical expertise remains important, but organisations are also looking for people who know how to collaborate with AI systems, verify outputs, manage risks and make informed decisions.

This positions institutions such as Howest at the forefront of a broader educational transition. The most future-ready institutions will not be those that simply allow or prohibit AI. They will be those that explicitly teach students how to use AI professionally, responsibly and critically.

In many ways, AI literacy is becoming what digital literacy was twenty years ago: not a specialist skill, but a foundational capability for every graduate entering the labour market.


3. UNESCO Calls for Human-Centred AI in Education

What happened?

At UNESCO’s Digital Learning Week, education ministers and representatives from more than 25 countries endorsed a statement calling for education to remain a public good in the age of AI. The declaration emphasises teacher agency, critical thinking and public accountability in AI deployment.
Source: UNESCO
https://www.unesco.org/en/articles/education-ministers-call-education-remain-common-good-age-ai-unescos-digital-learning-week

Why does it matter?

This is an important counterbalance to the increasingly commercial nature of educational AI. While technology providers are rapidly expanding their offerings, UNESCO is warning that educational systems must remain centred around human judgement and public values.

Why should Belgian readers care?

Belgium is already navigating questions around AI procurement, student data, academic integrity and assessment. UNESCO’s message is a reminder that schools and universities should not simply adopt AI because it exists. They need to decide how technology supports educational goals.


4. AI Literacy Is Becoming a Compliance Issue

What happened?

The EU AI Act continues moving from legislation to implementation. The European Commission highlighted this week that AI literacy obligations now form a practical part of AI governance and that national authorities have begun supervisory activities under the Act’s implementation timeline.
Source: European Commission AI Office
https://digital-strategy.ec.europa.eu/en/policies/ai-talent-skills-and-literacy

Why does it matter?

AI literacy is no longer just a training topic. It is becoming a governance requirement. Many organisations still approach AI training as an optional professional-development activity. European regulation increasingly views it as part of responsible AI deployment.

Why should Belgian readers care?

Public institutions, educational organisations and companies in Belgium should pay close attention. The question is no longer whether staff need AI training, but whether organisations can demonstrate that employees have adequate understanding of the systems they use.


5. Europe’s Digital Skills Agenda Gains Momentum

What happened?

The European Commission confirmed that findings from its new education reports will feed into its forthcoming Education Package and broader Union of Skills initiative. The focus includes digital skills, teacher development and educational infrastructure.
Source: European Commission
https://education.ec.europa.eu/whats-new/news/new-reports-examine-how-ai-and-digital-technologies-are-shaping-education-in-europe

Why does it matter?

Skills policy is increasingly becoming economic policy. Europe faces simultaneous shortages in digital, engineering, cybersecurity and AI-related roles. Policymakers are beginning to treat skills development as strategic infrastructure.

Why should Belgian readers care?

Belgium’s competitiveness increasingly depends on talent availability. Whether you’re in cybersecurity, manufacturing, healthcare, software development or research, talent shortages remain one of the most immediate constraints on growth and innovation.


6. Sovereign AI Moves from Concept to Infrastructure

What happened?

This week saw significant investment announcements around sovereign AI infrastructure. One notable example was a six-country rollout of sovereign AI platforms built on NVIDIA technologies, including national-scale AI infrastructure plans and locally hosted models.
Source: MeetKai / NVIDIA ecosystem announcement
https://www.prnewswire.com/news-releases/meetkai-takes-sovereign-ai-built-on-nvidia-to-six-countries-302890312.html

Why does it matter?

The conversation around AI is increasingly shifting from models to infrastructure.

Countries are recognising that access to compute, data centres and local AI capabilities may become as strategically important as energy or telecommunications infrastructure.

Why should Belgian readers care?

Europe continues to debate digital sovereignty, cloud dependence and technological resilience. The same discussion is relevant for Belgium. Questions about where data is processed, who controls AI infrastructure and how local innovation ecosystems develop will only become more important.


7. AI Infrastructure Spending Continues to Accelerate

What happened?

New industry analysis highlighted the enormous scale of investment flowing into AI infrastructure globally. Estimates cited this week project hundreds of billions of dollars being invested in AI compute and supporting infrastructure during 2026.
Source: McKinsey Technology Trends coverage
https://cfo.economictimes.indiatimes.com/news/cfo-tech/ai-infrastructure-investment-seen-surging-to-769-billion-in-2026-amid-growing-safety-concerns/134484013

Why does it matter?

Much of the public discussion focuses on chatbots and AI applications. Behind the scenes, the real story may be the unprecedented construction of computational infrastructure.

These investments influence research capacity, industrial competitiveness, startup ecosystems and labour markets.

Why should Belgian readers care?

European universities, research centres and innovative companies compete within a global innovation system. Access to advanced compute resources increasingly affects who can conduct cutting-edge research and build next-generation products.


Three Trends Worth Watching

1. AI is becoming an institutional challenge

The week’s developments were not primarily about new AI capabilities. They were about governance, skills, teacher training and organisational readiness.

The question is increasingly how institutions adapt.

2. Skills are moving to the centre of policy

The European Commission, UNESCO, universities and employers are converging on a shared concern: skills.

AI literacy, digital competence and continuous reskilling are becoming strategic priorities across sectors.

3. Infrastructure is becoming the new battleground

Just as cloud computing became a strategic capability during the past decade, AI infrastructure is becoming a strategic asset today.

Countries, research institutions and companies are competing to secure access to compute, talent and data ecosystems.


The most interesting development this week was not a new AI model or a flashy product launch. It was the growing recognition that technology alone does not determine outcomes. Skills, governance, education and institutional choices matter just as much.

In many ways, the next phase of the AI era may be less about what the technology can do and more about what societies decide to do with it. Belgium, with its strong research base, international outlook and education sector, is well positioned to contribute to that conversation. The real question is whether we move quickly enough from awareness to action.

What Belgium should know about this week’s AI, education and innovation developments

If you only have ten minutes this weekend to catch up on technology, education and innovation news, these are the developments I would focus on.

What struck me this week is that several seemingly separate stories point in the same direction. Artificial intelligence is no longer mainly about experimentation. Governments, universities and companies are increasingly dealing with questions of governance, skills, adoption and public trust. For Belgium, that may be even more important than the latest model release.


1. UNESCO puts AI governance in education at the centre of the global agenda

What happened?

During UNESCO’s Digital Learning Week 2026 in Paris, education ministers and representatives from more than 25 countries adopted a joint statement on the future of education in the age of AI. The statement says that AI should support learning rather than replace human judgement, teaching and critical thinking. UNESCO also launched a global consultation on AI governance in education.

Why does it matter?

The debate has shifted. A year ago, many discussions focused on whether AI should be allowed in classrooms. Today the question is how educational systems can remain in control while using the technology productively.

UNESCO’s position is clear: AI may help learners and teachers, but educational decisions should remain grounded in human agency, public values and transparency.

Why should Belgian readers care?

Belgian schools, colleges and universities are all wrestling with AI policies. UNESCO’s recommendations closely align with debates already taking place across Flemish and federal education networks: assessment redesign, AI literacy, teacher training and academic integrity.

For education leaders, this is another signal that AI governance is becoming a strategic leadership responsibility rather than an IT issue.

Sources


2. Belgium updates its national AI strategy

What happened?

Belgium adopted an updated national artificial intelligence strategy this week. According to reports, the renewed strategy focuses on AI skills, research, implementation, public services and maintaining strategic control over AI development and deployment.

Why does it matter?

Many countries have AI strategies. Fewer are moving from policy documents to implementation and coordination.

The revised Belgian approach appears intended to accelerate AI adoption while strengthening sovereignty, governance and skills development. It reflects a wider European concern: how to benefit from global AI innovation without becoming entirely dependent on external technology providers.

Why should Belgian readers care?

Belgia already has strong AI research clusters, universities, startups and industrial expertise. The real challenge is scaling adoption. For businesses, this may create new opportunities for funding, partnerships and innovation projects. For higher education, it reinforces the importance of AI literacy and workforce development.

Sources


3. The EU’s AI Act moves further into its implementation phase

What happened?

Europe continues entering the operational phase of the AI Act. New obligations for providers of general-purpose AI models have become enforceable, while preparations continue for more extensive obligations affecting high-risk AI applications.

Why does it matter?

The AI Act is no longer just a policy discussion.

Organisations increasingly need practical answers to questions such as:

  • How are AI systems documented?
  • How is transparency ensured?
  • What evidence is required for compliance?
  • How are risks monitored?

The focus is moving from regulation design to implementation capability.

Why should Belgian readers care?

Education, employment and public services are among the sectors explicitly referenced within the AI Act framework. Universities, school groups, HR departments, startups and public administrations that deploy AI systems will increasingly have to understand regulatory obligations. This creates growing demand for legal, technical and governance expertise.

Sources


4. Europe doubles down on skills as a competitiveness strategy

What happened?

The European Commission continued positioning the “Union of Skills” as a central element of Europe’s competitiveness agenda. Recent European communications and education initiatives emphasise STEM education, microcredentials, lifelong learning and labour market mobility.

Why does it matter?

For years Europe worried about digital infrastructure. Now it is increasingly worried about people. Across industries, employers report shortages in digital, engineering, cybersecurity and AI-related skills. Europe’s response is becoming clearer: invest heavily in education, reskilling and workforce mobility.

Why should Belgian readers care?

Belgium faces similar challenges.

Companies struggle to recruit technical talent. Universities face pressure to prepare students for rapidly changing careers. Governments must balance economic competitiveness with social inclusion. The European skills agenda will increasingly influence funding programmes, curriculum reforms and workforce policies.

Sources


5. Microsoft opens discussion on long-term AI governance

What happened?

Microsoft released a draft “Humanist AI Code of Conduct” and launched a public consultation process around future AI governance principles. The document includes commitments related to human oversight, transparency and preventing harmful uses of advanced AI systems.

Why does it matter?

The most interesting aspect is not the document itself. It is the fact that one of the world’s largest AI players is openly discussing governance structures before future generations of AI systems arrive. That suggests the conversation is expanding beyond performance benchmarks and product announcements.

Why should Belgian readers care?

Belgian organisations increasingly rely on global AI platforms.

The governance decisions made by companies such as Microsoft, OpenAI, Google and Anthropic will influence classroom tools, business applications and public-sector systems used throughout Europe. Understanding these governance approaches is becoming as important as understanding the technology itself.

Sources


6. UNESCO highlights a challenge many institutions recognise: moving beyond AI pilots

What happened?

A UNESCO strategy lab this week focused on a surprisingly practical question: how can AI projects move beyond successful pilot phases and become sustainable at scale? Participants explored educator readiness, infrastructure, governance and long-term adoption challenges.

Why does it matter?

Many organisations have already proven that AI can work. Far fewer have proven they can adopt it sustainably. The biggest barriers increasingly involve organisational capability rather than technology.

Why should Belgian readers care?

Whether in a university, a school board, a healthcare organisation or an SME, the challenge is often the same:

  • How do we train staff?
  • How do we redesign processes?
  • How do we measure value?
  • How do we govern usage?

The institutions that solve those questions are likely to benefit the most from AI in the coming years.

Sources


Three Trends Worth Watching

1. AI is becoming a governance challenge

Several stories this week point in the same direction.

Whether it is UNESCO, the European Union, Belgium or Microsoft, discussions increasingly focus on accountability, oversight, transparency and public trust.

The technology race continues, but governance is catching up.

2. Skills are becoming strategic infrastructure

For years governments invested in roads, broadband and research facilities. Today skills are increasingly treated the same way. The ability to retrain workers, educate students and build AI literacy is becoming a competitiveness issue.

3. Institutions are moving from experimentation to implementation

The biggest question is no longer whether AI works. The question is whether organisations can embed it effectively into teaching, research, business processes and public services. That requires leadership, policy and organisational change.


What This Means for Belgium

For educators

Expect growing emphasis on AI literacy, assessment redesign and teacher professionalisation. The challenge is not simply teaching students to use AI, but teaching them when to trust it, question it and improve upon it. At Howest I teach an AI Literacy course btw!

For businesses

The biggest competitive advantage may come from workforce capability rather than model capability. Organisations that combine AI tools with strong human expertise will likely outperform those that simply deploy technology.

For policymakers

The focus should increasingly move from awareness to implementation. Belgium has strong foundations in research and innovation. The next challenge is scaling adoption while maintaining public trust.

For universities and research centres

Opportunities remain significant. European funding priorities continue to align strongly with AI, digital transformation, education innovation and international collaboration.

Institutions capable of connecting these domains will be well positioned.


One theme kept returning this week: control.

Not control in the sense of restricting innovation, but in the sense of ensuring that institutions remain capable of shaping it.

AI is becoming more powerful every month, but the longer-term question is not how intelligent our systems become. It is whether our schools, companies, governments and universities become equally capable of making informed decisions about how those systems are used. For Belgium, that may turn out to be the most important innovation challenge of all.

AI zal niet het verschil maken tussen bedrijven. AI-geletterdheid wel.

De afgelopen week verschenen twee stukken die me aan het denken zetten.

In een nieuw essay waarschuwt Bill Gates dat artificiële intelligentie zowel de grootste gelijkmaker ooit als de grootste bron van ongelijkheid kan worden. Volgens hem onderschatten veel mensen nog steeds hoe ingrijpend de impact van AI zal zijn op ons werk, onze economie en onze samenleving.

In een interview voor ICT Magazine sprak ik over een andere vaststelling: AI verandert de job van IT’ers fundamenteel. Niet omdat AI mensen vervangt, maar omdat de aard van het werk verschuift. Minder uitvoeren, meer analyseren, beoordelen en orkestreren.

Op het eerste gezicht lijken dit twee verschillende verhalen. Dat zijn ze niet.

De verkeerde vraag

Wanneer organisaties over AI praten, gaat het vaak over technologie.

  • Welk model gebruiken we?
  • Welke tools moeten we aankopen?
  • Hoe snel kunnen we automatiseren?
  • Hoeveel productiviteitswinst kunnen we behalen?

Dat zijn relevante vragen. Maar ze zijn niet de belangrijkste. De vraag die volgens mij veel meer aandacht verdient is: hebben onze medewerkers voldoende AI-geletterdheid om deze technologie verstandig, kritisch en doelgericht te gebruiken? Want toegang tot AI wordt steeds minder een concurrentieel voordeel. Vandaag kan een multinational dezelfde generatie AI-tools gebruiken als een kleine KMO. Het verschil zit niet langer in toegang, maar in gebruik.

Iedereen heeft AI. Niet iedereen heeft AI-vaardigheden.

In het ICT Magazine-artikel wordt AI-kennis omschreven als een nieuwe basiscompetentie voor IT-profielen. Niet alleen voor ontwikkelaars, maar ook voor architecten, consultants en securityspecialisten. Ik geloof dat dit zelfs breder gaat. Binnen enkele jaren zal AI-geletterdheid even vanzelfsprekend zijn als digitale geletterdheid vandaag. Niet iedereen moet een AI-expert worden. Maar iedereen zal moeten begrijpen:

  • wat AI goed kan;
  • waar AI fouten maakt;
  • hoe AI-systemen tot antwoorden komen;
  • welke risico’s verbonden zijn aan AI-gebruik;
  • wanneer menselijke controle noodzakelijk blijft.

Dat is niet langer een nice-to-have. Dat is digitale basisvaardigheid.

Prompting is niet hetzelfde als AI-geletterdheid

Vandaag zien we vaak dat organisaties medewerkers een workshop prompting geven en vervolgens denken dat de AI-transformatie gestart is. Dat is begrijpelijk, maar onvoldoende. Goede prompts schrijven is nuttig. Zie ook mijn boek daarover 😉 Maar AI-geletterdheid gaat over veel meer:

  • kritisch denken;
  • informatie beoordelen;
  • broncontrole;
  • ethisch gebruik;
  • gegevensbescherming;
  • begrip van beperkingen en bias;
  • weten wanneer je AI wel of niet inzet.

Iemand die een indrukwekkende prompt kan schrijven maar elk AI-antwoord blind vertrouwt, is niet AI-geletterd. Net zoals iemand die een zoekmachine kan gebruiken nog geen onderzoeker is.

De nieuwe digitale kloof

Bill Gates waarschuwt dat AI ongelijkheid kan vergroten als we niet tijdig ingrijpen. Ik denk dat die kloof niet alleen tussen landen of bedrijven zal ontstaan. Ze zal ook ontstaan tussen mensen. Tussen mensen die AI leren inzetten als versterking van hun expertise en mensen die AI blijven zien als een magische antwoordmachine. De eerste groep zal sneller leren, betere beslissingen nemen en productiever worden. De tweede groep riskeert afhankelijk te worden van technologie die ze onvoldoende begrijpt. Daarom is onderwijs zo belangrijk. Maar ook bedrijven dragen hierin verantwoordelijkheid. AI-geletterdheid mag geen exclusieve kennis worden van enkele enthousiastelingen of innovatie-experts. Ze moet deel worden van de basisvorming van elke medewerker.

Wat moeten we vandaag doen?

Volgens mij zijn er drie prioriteiten.

1. Maak AI-geletterdheid toegankelijk voor iedereen.

Niet alleen voor technische profielen, maar ook voor HR, finance, marketing, administratie en management.

2. Leer mensen kritisch omgaan met AI-output.

De belangrijkste vaardigheid van morgen is misschien niet het genereren van antwoorden, maar het beoordelen ervan.

3. Focus op menselijk inzicht.

Naarmate AI beter wordt in uitvoering, wordt menselijk begrip waardevoller. Technologie kan antwoorden genereren. Mensen moeten nog steeds de juiste vragen stellen.

Slot

In het ICT Magazine-interview gebruikte ik de uitspraak:“Syntax is gratis geworden, maar begrip is schaars.” Misschien vat dat ook de bredere AI-discussie samen. De komende jaren zullen steeds meer mensen toegang krijgen tot krachtige AI-systemen. Dat is niet de uitdaging. De uitdaging is ervoor zorgen dat voldoende mensen begrijpen hoe ze die systemen verstandig, kritisch en verantwoord gebruiken. AI zal waarschijnlijk niet het verschil maken tussen succesvolle en minder succesvolle organisaties.

AI-geletterdheid wel.

De toekomst vraagt om meer dan technische kennis alleen. Daarom ontwikkelden we bij Howest, samen met mijn collega Tibo Van Craenenbroeck, een opleidingsmodule rond AI-geletterdheid. We willen onze studenten niet enkel leren werken met AI, maar hen ook inzicht geven in de kansen, beperkingen en maatschappelijke impact ervan.

Zo maken we hen klaar voor een arbeidsmarkt waarin AI geen uitzondering meer is, maar een vanzelfsprekend onderdeel van het dagelijkse werk.