Jolyce Demely

The through-line

One argument runs through everything: Flanders cannot compete on price, so it has to compete on innovation, and that makes AI literacy the single competence that decides who stays in the game. "We gaan hier nooit de laatste energiekost hebben. We gaan nooit de laatste loonkost hebben. Het enige waar dat wij het verschil kunnen op maken, dat is net in die innovatie" [1][2]. The corollary is put bluntly at the level of the individual: "Wie niet mee is met die AI sneltrein, ja, die dreigt eigenlijk wel uit de boot te vallen" [1][2], and at the level of the region: "Er niet mee aan de slag gaan gaat ervoor zorgen dat we achterblijven" [2].

The second half of the argument is that the response has to come from education and business together, not from either alone, and that both are moving too slowly for how fast the ground is shifting. The pace is itself part of the case: the same study run a year earlier would have produced different results [2], and curriculum end goals written over the summer holidays are already out of date by the end of the school year [1][2].

On AI literacy as the new dividing line

The familiar split between highly-educated and low-educated workers is disappearing, and what replaces it is whether you are AI-literate or not [1][2]. Alongside that, manual trades are being revalued rather than devalued [1][2]. The framing is deliberately non-hierarchical: the question is no longer diploma level but whether you can work with the technology.

Refusing the technology is not treated as a defensible position. "Ik hou niet zo van die aversie van technologie. Dat is eh ja, de toekomst een stukje ontkennen" [1][2].

On what AI does to software teams

The most concrete finding is a reversal inside IT firms: developers now spend only a quarter of their time on the software itself. "Eigenlijk zijn ze nog 25% van de tijd bezig met het ontwikkelen van de software, maar 75% van de tijd gaat vooral naar de klant" [1][2]. That remaining three quarters goes to client contact, understanding the problem and validating output, which means the bottleneck has moved to the user side, to whether the client can keep up [1][2].

Team structure is changing with it. Dedicated AI departments inside IT companies are becoming obsolete as AI expertise is spread transversally into smaller, hybrid, multidisciplinary teams, and that is reshaping company governance quickly [1][2]. The make-or-buy line is moving too: companies will build small AI tools themselves, while IT firms get reserved for large projects involving critical data and integration needs, even as total demand for software keeps growing exponentially [2].

On why manufacturing is not moving fast enough

Flemish manufacturers are mostly doing the low-hanging fruit, copilots and small projects, while the large disruptive process-optimisation work stays absent [1][2]. That is measured against a competitive reality: Chinese dark factories [1][2]. Part of the explanation is capacity rather than unwillingness, since many SMEs are too busy surviving to invest in the shift [2], but the conclusion is that the reactive mindset has to be flipped [1].

On rebuilding curricula with slack in them

The proposal for schools is structural: fix only 70% of the curriculum and leave 30% free for innovation, experimentation, hackathons and projects on the company floor [1][2]. The reasoning is about obsolescence, since goals set in summer no longer hold by year's end [1][2].

Assessment should change in the same direction. "We gaan ze toch niet met AI leren hun huiswerk maken. Ja, absoluut wel. En vragen hoe ze geprompt hebben, waar ze bijgestuurd hebben en beoordeel ze vooral daarop en niet zozeer op het eindresultaat alleen" [2]. Banning technology from classrooms solves nothing, because it is everywhere on the work floor anyway [1].

On business as the biggest school of Flanders

"Ik denk dat de bedrijfswereld wel eens de grootste school van Vlaanderen zou moeten worden" [1][2]. In practice that means workplace learning inside every study programme, political science included, starting from age sixteen, organised as triple-helix collaboration between education, IT companies and industry [1].

This connects to a specific problem with young graduates: clients demand a maturity that fresh graduates do not have, so a jongeren strategy is needed in which business experience is built early through project work during education [2].

On 'shift up or shift left'

Young people should still study IT [2], but the path forks. Either shift up, deepening into hardcore programming specialisation, where fewer people will be needed, or shift left, broadening into business models and human skills [1][2]. The framing follows directly from the 25/75 split in how developer time is actually spent [1][2].

Takeaways

• Competitiveness will not come from cost. Belgium and Flanders will never have the lowest energy or labour costs, so innovation is the only remaining lever [1][2].
• The relevant skills gap is no longer high-educated versus low-educated but AI-literate versus not, and manual work is being revalued in the process [1][2].
• Developers at IT firms now spend 25% of their time building software and 75% on the client, so the bottleneck has moved to whether the user can keep up [1][2].
• Dedicated AI departments are on the way out; expertise is being distributed transversally into smaller multidisciplinary teams, which changes governance quickly [1][2].
• Fix 70% of the school curriculum and leave 30% open for hackathons, experimentation and company-floor projects, because annual end goals are outdated within the year [1][2].
• Let students use AI for homework and grade them on how they prompted and corrected it, not on the end result alone [1][2].
• Put workplace learning from age 16 into every study programme, political science included, through triple-helix collaboration between education, IT firms and industry [1].
• IT students face a choice: shift up into deep programming specialisation, where fewer are needed, or shift left into business models and human skills [1][2].