AI Can Make Mistakes: Why Media Literacy Matters More Than Ever

In a world where artificial intelligence can create flawless headlines, realistic images, and copy human voices with amazing accuracy, telling what's real from what's not has become a daily challenge. Consider a situation that's becoming more common: a convincing article appears online, complete with expert quotes, specific dates, and professional language. It spreads across platforms, shaping opinions, and informing decisions until it’s revealed that the quotes were fake, the experts never existed, and the story was entirely created by AI. 

This is our reality today, where the line between real information and artificial content grows harder to see. Generative AI arrived with immense promise, able to process vast data and deliver information faster than ever. But there's a problem and it's a serious one. 

Large language models (LLMs) have an odd habit: they "hallucinate." They create information that sounds believable but is completely made up like fake quotes, invented sources, false statistics. Here's the worrying part: these errors aren't rare mistakes. They're built into how these systems work. AI models are designed to be good at answering questions. When they're unsure, they guess because guessing actually helps their performance. The result? Wrong answers delivered with complete confidence. 

The danger isn’t simply that AI can make mistakes, but that these mistakes can mislead citizens, sway leaders, and shape public opinion — all while appearing entirely credible.

This is why UNESCO has launched its "AI Can Make Mistakes" campaign during Global Media and Information Literacy (MIL) Week 2025. The message is clear: in a world increasingly shaped by artificial intelligence, our most powerful tool is still our ability to think critically about information. 

Media and Information Literacy is not just an academic term. It’s an essential life skill. We can all take small but powerful actions: pause before sharing, question the source, diversify where we get information, and learn how AI systems work. These habits slow the spread of false information and rebuild trust. 

But lasting progress requires more than individual effort. Digital platforms should integrate Media and Information Literacy into their design. AI developers should be transparent about how their systems operate. UNESCO’s Guidelines for the Governance of Digital Platforms and its companion document, the Multistakeholder Action Plan for Integrating Media and Information Literacy on Digital Platforms offer a roadmap for strengthening users’ critical thinking skills and ensuring safer, more ethical digital environments. 

The global picture underscores the urgency. While 88% of countries recognize MIL’s importance in national policies, only 9% of countries have adopted standalone MIL policies. There is still much work ahead. 

This year's Global MIL Week, running from October 24-31 under the theme "Minds Over AI – MIL in Digital Spaces," brings this conversation to Colombia for its main conference. The goal is to explore how stronger Media and Information Literacy can help people engage critically with AI-created content. 

Join the conversation at #GlobalMILWeek and learn more about staying media and information literate on our website.