Learning after COVID-19 and the ICT career aspirations: Are students entering the AI era with weaker skills?

URL SCAN:
Learning after COVID-19 and the ICT career aspirations: Are students entering the AI era with weaker skills?

FIRST LINE:
Computer Science > Computers and Society [Submitted on 17 Apr 2026]

THE DISSECTION

This paper is institutional anxiety performing empirical rigor. Researchers examining whether the education system — which they are institutionally invested in — adequately prepares students for an AI-driven labor market, using PISA 2018/2022 data and a Variational Autoencoder. The entire framework is structurally backwards: measuring whether human skill development is sufficient for a labor market that DT logic proves will not require human cognitive labor at scale.

THE CORE FALLACY

Treating "skills" as the relevant variable when the mechanism is structural displacement.

The paper finds digital skills are "the strongest and most consistent predictor" of ICT career aspirations. This is precisely the kill mechanism dressed as a finding. The DT axiom is not that students lack skills — it is that AI severs the connection between skills, employment, and consumption regardless of skill level. Students entering with stronger digital skills are not better positioned; they are entering a domain AI will dominate faster, displacing them from the very work their aspirations target.

The paper uses "AI era" in its title while measuring readiness through PISA metrics designed for the 2018 labor market. It cannot see what it is studying.

HIDDEN ASSUMPTIONS

1. The ICT career pipeline remains human-viable. The entire dependent variable — ICT career aspirations — assumes these are desirable targets. DT says they are the first to be automated.
2. Teacher support and autonomy are structurally relevant. These are lag-domain inputs measuring satisfaction and engagement, not displacement resistance.
3. PISA 2018-2022 captures the relevant discontinuity. The paper literally ends its data window in 2022. GPT-4 was March 2023. The discontinuity the thesis describes falls entirely outside the data.
4. Country-level aggregated data can resolve individual displacement risk. A student in Singapore with "strong digital skills" faces the same AI displacement logic as one in Indonesia. The clustering analysis produces geographic groupings of students equally doomed by different national circumstances.

SOCIAL FUNCTION

Institutional self-exoneration theater. This is education researchers generating evidence that their institution matters. The policy recommendations — improve teacher support, enhance autonomy, strengthen digital skills — are what the education system can do. They are also, under DT logic, insufficient by definition. The paper provides empirical cover for educational reformism at the precise moment reformism becomes structurally irrelevant.

Partial truth wrapped in prestige formatting. Yes, the PISA data shows something real about skill distributions. No, the interpretation of those distributions as preparatory is a category error.

THE VERDICT

This paper is a lag-document from inside the collapsing interval. Published April 2026, analyzing data through 2022, using metrics validated in a labor market that will not exist by 2030 in any recognizable form. The ICT career aspirations it measures as dependent variables are themselves subject to AI-mediated obsolescence before the students holding them graduate. The paper's integration of "learning loss, digital skills and career expectations" treats three independently dying metrics as components of a single coherent problem. They are not. They are separate symptoms of the same structural discontinuity.

The researchers are not wrong that the education system is failing. They are wrong that the failure is correctable within the education system. DT is the theoretical framework this paper needs and cannot use because its authors are institutionally positioned inside the structure being diagnosed.

Viability: The paper's policy recommendations will be implemented, will show measurable effects on PISA-adjacent metrics, and will fail to alter the displacement trajectory. This is not a failure of the researchers. It is the logical endpoint of studying a transition system using frameworks native to the system being transitioned from.