GCRI · Campaigns
Responsible AI literacy
Artificial intelligence now shapes what people read, which services they receive and how content is made. Responsible AI literacy is a practical understanding of what these systems do, where they fail and what rights people have when they are used. It covers recognising synthetic media, judging generated answers and knowing when automated decisions can be questioned. This page sets out how misunderstanding of AI forms and spreads, how literacy is measured, and the decisions held by educators, regulators and platforms.
Responsible AI literacy on the record.
Responsible AI literacy: how the risk develops.
Practical understanding of AI risks. The dossier below sets out how the risk develops, what is measured, where the data come from, how it is modelled and who decides.
Capability is misjudged
People over-trust fluent machine output, a pattern known as automation bias, and sometimes dismiss useful tools after a single failure. Conversational systems invite people to attribute understanding and intent they do not have, which shapes how carefully their answers are checked.
Provenance blurs
Synthetic images, audio and text can be produced cheaply and are hard to tell apart from authentic material. The doubt they create also lets genuine evidence be dismissed as fabricated, a pattern described as the liar’s dividend.
Harms fall unevenly
Errors and bias in automated systems fall most on people least able to notice or contest them, such as benefit claimants, job applicants and speakers of less-resourced languages. Literacy work that reaches only confident users widens that gap.
Use outpaces understanding
Tools are adopted at work, in schools and in public services before users, teachers or public servants share a vocabulary for their limits. Misconceptions spread through peer advice and marketing claims and settle into practice before guidance exists.
A threshold model of collective behaviour.
Each agent acts once enough of its contacts have; small changes in the mean threshold or the early movers decide whether the behaviour stays local or cascades through the population. Teaching model with stated illustrative values, not a prediction.
Controls
- Mean threshold (share of neighbours)
- 18%
- Spread of thresholds
- 8 points
- Early movers
- 2.0% of people
- Contacts per person
- 8
Readouts
- Share who acted in the end
- 100%
- Global cascade
- Yes
- Steps to settle
- 8
Numbers, methods and sources
| Step | Share who have acted (%) |
|---|---|
| 0 | 3.3 |
| 1 | 10.0 |
| 2 | 26.3 |
| 3 | 49.3 |
| 4 | 71.0 |
| 5 | 88.5 |
| 6 | 98.8 |
| 7 | 100.0 |
| 8 | 100.0 |
Methods
- Threshold models of collective behaviour Granovetter (1978), American Journal of Sociology 83
- Global cascades on random networks Watts (2002), PNAS 99
- Complex contagions Centola and Macy (2007), American Journal of Sociology 113
- One-at-a-time sensitivity (tornado) Saltelli and others (2008), Global Sensitivity Analysis: The Primer, Wiley
Data
No observed data enters this model: it runs on the teaching values set by the controls, to be replaced with a programme’s own evidence.
Across the six pillars, led by Nexus Consortium
- Experiments Labs
- Built in Campaigns in Foundry
- Tested at Campaigns in Universe
- Codified in Governance, Security and Society in Standards
- Published in Campaigns in Reports
- Taught in Faculty of Governance, Policy and International Affairs in Academy
AI literacy matters wherever AI enters a domain’s services.
Findings are designed for the Exponential Technology series. UNESCO’s Recommendation on the Ethics of AI (2021) includes AI literacy among its policy areas.
- Curricula and evaluations.
- Accounts of AI tools in daily life.
How Responsible AI literacy is modelled, measured and tested.
The model, its assumptions, its data and its uncertainty, in the form a scientist or engineer would check. Structures follow published methods. The values drawn are illustrative, not findings.
System model
The couplings between the systems in Responsible AI literacy: signs of influence, a reinforcing and a balancing loop, and cross-links where stress jumps.
What is measured, and how well.
Each indicator carries its unit, its cadence, the kind of source behind it and the evidence label a value would carry. No values are shown: a programme supplies them under its own authority.
| Indicator | Unit | Cadence | Source type | Label |
|---|---|---|---|---|
| Self-assessed understanding of AI | percent of respondents | per survey wave | Eurobarometer and national surveys | REPORTED |
| Accuracy in identifying synthetic media | percent correct in a controlled task | per study | Experimental audience research | OBSERVED |
| Awareness of rights over automated decisions | percent of respondents | per survey wave | National and EU surveys | REPORTED |
| Concern about telling real from fake content online | percent of respondents | annual | Reuters Institute Digital News Report | REPORTED |
| Teachers trained in the use of AI | percent of teachers | per survey cycle | OECD TALIS and education ministries | Published statistics |
| Reported AI incidents involving deception or impersonation | incidents per month | monthly | AI Incident Database | REPORTED |
| AI literacy measures in national policy | count of measures by type | annual | OECD.AI policy database | DECLARED |
Where the evidence comes from.
Named public datasets and institutional records, each with its licence class. Sources by agreement are used only under their terms and never republished.
| Source | What it gives | Licence class |
|---|---|---|
| Eurobarometer surveys on artificial intelligence and the digital decade | Awareness, attitudes and expectations of AI across EU member states | Free reuse with attribution |
| Stanford AI Index | Annual compilation of AI research, deployment, policy and public opinion data | CC BY-ND 4.0 |
| OECD.AI Policy Observatory | National AI strategies and policy measures, including education and skills | OECD terms, free with attribution |
| AI Incident Database | Reported incidents of harm or near harm involving AI systems | CC BY-SA 4.0 |
| Reuters Institute Digital News Report | Public comfort with AI in news and concern about false content | Free to read; survey data by request |
| OECD TALIS | Teacher training, practice and use of digital tools | OECD terms, free with attribution |
| Pew Research Center surveys on AI | Public awareness of and attitudes to AI in several countries | Free to use with attribution; datasets on request |
Who decides, and what the evidence serves.
Campaigns prepares evidence. The decisions below are taken by the authorities named, under their own rules, and never by the platform.
| Who decides | The question | Evidence this page serves |
|---|---|---|
| Ministries of education and curriculum authorities | Which AI competencies students and teachers need, and at which stages | Teacher training data; competency frameworks; tested teaching material |
| AI and data protection regulators | How to supervise AI literacy and transparency obligations, such as Article 4 of the EU AI Act | Awareness surveys; incident records; literacy measures in policy |
| Online platforms and AI providers | How to label synthetic content and explain system limits to users | Research on label comprehension; incident patterns; provenance standards |
| Electoral management bodies | How to prepare voters for synthetic media during elections | Detection task research; rumour logs; trust in electoral information |
Outputs this topic produces
- Practical explainers on how generative and predictive systems work and fail
- Exercises on checking source and provenance of media, with their limits stated
- Plain-language guides to rights over automated decisions under applicable law
- Teaching material for educators aligned with UNESCO AI competency frameworks
What experts ask.
- 01
What should a non-specialist understand?
That AI systems learn patterns from data, can be confidently wrong and reflect the limits of what they were trained on.
- 02
Which risks are practical today?
Fabricated content, impersonation, automated errors in services, privacy loss and over-reliance on generated answers.
- 03
How can people check what they see?
Habits of verification: source tracing, provenance signals and asking who benefits from a claim.
- 04
What recourse exists?
How people can question an automated decision, correct data about themselves or report misuse.
What drives the outcome.
Generative models produce fluent text, images and audio by predicting plausible outputs, not by checking facts. Fluency invites trust, which is why errors and fabrications can spread. Synthetic media lowers the cost of impersonation and of producing misleading material at scale.
Automated decision systems used in lending, hiring, benefits or policing can embed historical patterns in their data, and their reasoning may be hard to explain. Errors fall unevenly, and people affected may not know an automated system was involved.
Public discussion swings between alarm and hype. Both distort judgement. Literacy work aims for calibrated understanding: what current systems do reliably, where they fail and which safeguards exist.
How evidence is gathered, and where it stops.
| Method | What it shows | Its limit |
|---|---|---|
| Misconception surveys | Common beliefs about how AI works | Beliefs shift quickly with new tools |
| Hands-on exercises | Whether people can spot errors or synthetic content | Detection skills may not transfer to new formats |
| Expert review | Accuracy of claims about capabilities | Capabilities change between review and reading |
| Follow-up measures | Whether verification habits persist | Attrition and self-report bias |
How it is studied, and what the methods cannot say.
Where this page’s references lead.
The bodies and documents this page is read against. Status and editions change; each link goes to the issuing body.
Questions and answers.
Can people learn to spot deepfakes?
What is content provenance?
What rights do people have over automated decisions?
Does Campaigns rate AI products?
Does the platform evaluate or rate AI products?
Is AI used to produce campaign material?
How does the material stay current?
Contribute evidence from this page.
Reports is the Consortium’s quarterly publication. What you know about this subject can become evidence, a case or a correction in the next edition, credited to you and published under CC BY 4.0 where it is your own work.
- EmailGateway GW-GCRISend it from this pageEvidence, a correction, a case, a method, a review, a translation or profile input. It arrives tagged with this page, so the editor who owns it receives it.
- ZenodoOpen depositDeposit it and submit to the communityPublish your own record on Zenodo with your ORCID iD, then submit it to the Reports community there. Curators review it; the DOI stays yours.
- ReportsGCRISee every way to contributeThe contribution types, how each is reviewed, what credit you receive, the licence and what is never accepted.
Never send personal data, sensitive infrastructure detail or anything you have no right to share. A contribution is reviewed before use; sending it confers no authorship or editorial rights, and sponsorship never buys a finding.
Bring a method, a dataset or a question.
Information and enquiries are open today. An enquiry receives a reference and names the next step; applications, payments, compute allocations and registrations open only when their team, terms and review capacity exist.
Legal documents.
Maps, data and figures on this site are shown for context and general information. They are not advice, forecasts or warnings; the limits of the information are set out in the Disclaimers and Data Notices.