Schools 'desperate' for clearer AI guidance - Tes

TEXT START: Teachers are “desperate” for clearer guidance on how AI should be used in schools, MPs have been told.

The Dissection

The article documents institutional lag. AI is already entering classrooms, coursework, tutoring, marking, and administration, while schools lack rules for containing it. Its real subject is not guidance but the decomposition of educational work: generation becomes automated, checking becomes additional labor, and credentials become harder to trust.

It correctly records several early symptoms of the Discontinuity Thesis: AI does not automatically reduce workload; it can shift labor into verification and system management; disadvantaged schools are more exposed because human capacity is already scarce; and unsupervised written assessment is losing its ability to reveal individual competence.

But the article frames AI as a dangerous adjunct that can be made safe through training, statutory guidance, better implementation, and continued teacher relationships. That preserves the appearance that the institution remains in control.

The Core Fallacy

The central error is treating the crisis as a policy gap. The problem is not that schools lack a sufficiently clear manual. The problem is that AI changes the cost and speed of cognitive production, while rules cannot prevent diffusion when institutions, students, and employers all gain incentives to use it.

The reported checking burden is transitional overhead, not evidence that automation has failed. Early automation often creates supervision work before systems improve, standardize, and absorb more of the surrounding process. “Reconfiguring” teachers’ work can therefore be the bridge to reducing the amount of teacher labor required.

The article also mistakes current weaknesses in AI tutors for a permanent structural limit. Soulless systems and poor implementation are real constraints today, but they do not preserve human labor indefinitely. They are temporary friction.

Hidden Assumptions

The text assumes that:

• Human teachers will remain economically necessary if AI is properly governed.
• Training and rules can preserve a stable human-only domain at scale.
• AI will mainly supplement teachers rather than progressively replace portions of their work.
• High-quality teachers can be supplied to disadvantaged schools despite the shortages that make technological substitution attractive.
• Coursework can be redesigned without undermining the credibility of educational credentials.
• Purposeful, moderate AI use can remain stable once cheap, ubiquitous systems reward heavier use.
• The central problem is educational quality, rather than ownership and control of the systems performing the work.

The PISA evidence is also limited: the article explicitly states that lower scores among AI users do not establish causation. It shows a warning signal, not proof that AI necessarily reduces learning.

Social Function

This is partial truth serving transition management, with a layer of ideological anesthetic. It tells institutions how to absorb AI, retrain personnel, redesign assessments, and preserve legitimacy without confronting the larger consequence: productive participation becomes less necessary.

It is not pure propaganda. Its warnings about workload, inequality, cheating, and degraded learning are substantial. But by keeping the discussion inside classroom practice, it turns a structural labor displacement problem into a question of professional guidance and responsible use.

The Verdict

The article is a smoke alarm accurately describing smoke in the classroom while mistaking the building’s structural fire for a missing rulebook. Schools are not facing a temporary shortage of AI guidance; they are becoming a test site for the gradual automation of cognitive labor.

Teachers retain temporary servitor value in relationships, verification, accountability, and high-stakes judgment. Those are lag defenses, not a restoration of the old employment system. Guidance may delay institutional breakdown and protect assessment credibility for a time. It cannot reverse the underlying movement from mass human participation toward AI-controlled production.