Prompting critical thinking
Artificial intelligence can now produce text, images, and videos convincing enough to blur the line between authentic evidence and fabricated reality, making it harder for readers to know what to trust. As AI-generated content spreads across news feeds, social media, and everyday conversations, media literacy has become essential to recognizing misleading information, checking sources, and making informed decisions.
Did you notice anything unusual about the paragraph above? If not, we have some news for you: it was generated by artificial intelligence. The question now is whether recognizing AI-generated content is just a challenge of the present or an essential skill for the future.
In less than four years since the release of ChatGPT to the general public, about 50% of American adults have begun using artificial intelligence. Of these, 24% use AI on a daily basis. Microsoft estimated that 13.7% of the country’s working-age population used AI in the second half of 2025, while Coursera’s 2025 report recorded a 383% year-on-year increase in enrollments in AI courses.
Firsthand account
These rapidly developing trends can be observed across the globe, with Kazakhstan also seeing a growing interest in AI. Asiya Kaliyeva, 24, admits that she uses artificial intelligence on a daily basis for both work and personal purposes. Although she frequently uses artificial intelligence, Asiya admits that there have been instances where she mistook AI-generated content for the real thing.
“In recent years, there has been so much AI that I’m no longer able to distinguish between what is 100% human-made and what is AI-generated,” she says.
Asiya believes that AI-generated content falls into two categories: the first is the easiest to identify due to obvious origin – AI-generated images, films, or videos; the second category is more challenging for her to discern as it involves content created by AI in a less overt, hidden manner.
“I realize that texts can be written by AI – as can be scripts and ideas for short social media videos – but because it’s presented or released by humans, it doesn’t even cross my mind that it might have been AI-generated,” she shared.
Beyond the younger generation, older people are also embracing artificial intelligence. 78-year-old Alima Syzdykova has been using social media for over eight years, though she began using artificial intelligence quite recently. She believes that social media has completely replaced her need for entertainment.
“For me, social media has become an alternative to television,” she admits, adding that AI also became a tool for obtaining information. “If I don’t need to read through a forum or look at other people’s experiences, I just type my question into an AI and get the information I need. Most often, I ask about health or gardening.”
Despite the active usage, Syzdykova considers digital literacy to be especially important for herself, as she views herself as belonging to a potentially vulnerable category of users. The difference in users’ approach to AI lies in exactly that: while the younger generation strives to identify artificial intelligence, the older generation prefers not to trust what they see in the media.
“Fake information is everywhere these days – on television, on social media. I’ll continue to use these platforms, but I certainly don’t feel the need to take everything I see on them seriously,” she added.
Old tactics, new tools
Syzdykova’s caution reflects a much older problem: technologies change, but attempts to manipulate public opinion do not. Long before the rise of AI, propaganda exploited the most influential media of its time.
In Nazi Germany, newspapers, radio and films spread antisemitic messages and helped normalize the persecution of Jews. During the 1994 genocide against the Tutsi in Rwanda, the radio station RTLM broadcast anti-Tutsi propaganda and direct calls for violence. AI has not created these tactics, but it has introduced new ways to produce and spread manipulative content
Almabek Ibraiymov, a fact-checker and senior lecturer at Maqsut Narikbayev University, points to a shift from earlier forms of digital manipulation to the wider use of generative AI after 2022. These technologies did not create propaganda, but they have made manipulative content easier to produce and distribute on a larger scale.
The change is not limited to the creation of individual fake images or videos. Generative tools can support entire networks of misleading content, allowing large volumes of material to be produced, repeated and adapted for different audiences. Automated systems may then collect information from both reliable and fabricated sources, making particular narratives appear more widespread or credible than they are.
Tatyana Golubtsova, a media and information literacy expert and lecturer at Maqsut Narikbayev University, says such trends are already being observed in some countries. This can involve the deliberate creation of numerous websites containing propaganda, from which automated bots gather information and distribute it to different audiences.
AI can also present false claims in persuasive and accessible language, making them difficult to distinguish from reliable information. Before the widespread use of generative tools, only a select few were capable of creating convincing fakes. Now that such tools are widely available, trust in media information has been shaken up and is actively questioned.
Golubtsova identifies the loss of traceable sources as one of the most serious challenges AI poses for media literacy education.
“It might not even exist. It could be fabricated or generated. That is precisely why it represents one of the most serious threats,” she says.
Yet the fair conversation about AI is not entirely negative. Although AI-generated content can spread false or harmful information, it also offers opportunities for creative expression and entertainment.
“They create cool films using artificial intelligence and develop supportive AI-driven personas; while it’s clear during interaction that one is communicating with an AI, these personas feel relatable and are shaped by well-crafted prompts,” Golubtsova says.
Altered perception
In the B.A.I. (Editor’s note: the period Before Artificial Intelligence), people would search for information by entering a query and choosing from thousands of sourcesa process that required them to select and verify what they read.
Today, however, users often prefer to receive a ready-made answer from AI. It analyzes information quickly and extracts what appears to be relevant, though it may remain unclear whether that information comes from reliable sources. This makes the task easier, but it may also diminish critical thinking skills.
Golubtsova believes this convenience can come at a cost. “Our critical thinking declines, and we think the text is good and coherent. When working with students, I notice that they often feel this way. Therefore, it’s crucial to remain self-aware,” she says, adding that critical thinking remains “the foundation” for perceiving information.
Like any skill, it should be put into practice, not simply taught. Upon receiving a response from a neural network, the user should review the information provided, users need to identify any points that are open to debate, ask the network for links to its sources, verify those sources against other materials, and conduct an independent search to double-check the facts.
“For every questionable point or fact, we must identify a clear attribute confirming that it truly exists.”
She also believes that users should understand how to use AI responsibly. However, it is important to remember that people remain the most important factor.
“We cannot not use AI, as it is a technology we cannot do without. In many ways, it is genuinely helpful, allowing us to accomplish tasks much faster than we could have in the past,” she believes. “It is better if we write it ourselves – perhaps not as elegantly as an AI would, but it will be our own work. It is better if we do this task ourselves and take personal responsibility for it.”
After all, a network cannot assume responsibility. That responsibility always lies with the user who employs it and disseminates the results of its work.
Check, check, check again
Fact-checking is a skill everyone needs, regardless of age or profession. A useful starting point is to trace a claim to its original source, checking where it first appeared and what evidence supports it. For images, reverse image searches can help locate other appearances online and reveal the context in which they were published.
Verification also requires pausing before reacting to a message. “People often respond to information emotionally. They take it in without checking whether it is true,” Ibraiymov says. He recommends beginning with two basic questions: “Is this true? Where did it come from?”
But finding where a story began does not establish whether it is true. AI-generated content can describe events that never happened or attribute statements to people who never made them, Golubtsova warns: «We must verify whether what we see or read truly exists at all.”
The next step is to examine the message itself and the interests behind it. “When we receive information, we should first ask: ‘What does this information offer me? Who is behind it?’” he says. A claim should then be compared with independent sources rather than accepted on the basis of its first appearance.
Nor does seeing the same claim across multiple websites make it reliable. Repeated exposure can make information feel more credible simply because it becomes familiar — a phenomenon psychologists call the «illusory truth effect».
Research shows that repetition can increase belief even in false claims. The distinction is between repetition and independent confirmation. Several posts repeating the same unsupported claim offer no additional evidence, however widely they circulate.
Asking an AI to confirm a claim is not independent confirmation either, as there is also a question of people-pleasing or better described as user-pleasing. Despite the inclusion of the word “intelligence,” AI is not analogous to intelligence as it is classically understood.
“This is not intelligence in the conventional sense of the word, but merely a suitable set of neural data connected in various sequences. It is difficult for the program to grasp why a particular connection might be inconvenient for you, especially when it occurs so frequently.”
An important characteristic to be aware of is that a neural network does not possess its own point of view or personal stance. Instead, AI adapts its response and tone based on the user’s own opinions.
“Artificial intelligence is designed to support you; under no circumstances should it come into conflict with you, “ the expert confirms. “Not only does AI not argue with you — and may even go along with your mistakes — but it also continues to make its own errors.”
The future of critical thinking
Checking sources and questioning what we see online could become everyday habits, and according to Golubtsova, these skills should be taught from early childhood.
“This will be a natural skill. It will be embedded in our consciousness, and it should be developed as fully as possible from very early childhood,” she says.
For younger children, she suggests starting with familiar activities: stories, fairy tales and games that encourage them to ask questions and discuss what they hear. Children can also be encouraged to consider why they need a particular piece of information and how they might use it.
This emphasis on education is echoed by UNESCO, which advocates incorporating AI literacy into media and information literacy programmes. Its guidance highlights the need to assess source reliability and understand the different forms of evidence that AI can generate. Part of that literacy comes from practical experience. Golubtsova encourages people to work with AI directly as a way to understand how it works.
“To learn to understand something, we need to do it ourselves. That is how we grasp how technology works,” she says.
Recognizing AI-generated material, however, is only one part of evaluating information. Detection tools have limitations: the US National Institute of Standards and Technology (NIST) cautions that no individual technical approach provides a comprehensive solution for identifying and tracking synthetic content, and the effectiveness of such tools depends on the context in which they are used.
For readers, this means looking beyond whether an image or video appears convincing: comparing sources and considering whether the reported event or statement is plausible. Plausibility should be a starting point for checking, however, rather than proof. An unexpected statement is not necessarily fabricated, just as a believable one is not necessarily authentic.
Ultimately, it comes down to a broader habit of verification: tracing a claim to its original source, checking its context, and looking for independent evidence before sharing it. Ibraiymov offers a simple rule for putting that habit into practice:
“We should not rush to publish or share information as soon as we see it. First, we should check it, take some time and only then decide.”
To mark Global Media and Information Literacy Week, observed annually from 24 to 31 October, MNU Newsroom has compiled a checklist of essential steps for verifying information:
- Question what you receive. Ask yourself: Who sent this to me? Who created it? What does this message claim or confirm? What is the objective of those spreading this information?
- Always seek out the original source, rather than relying on reposts or summaries.
- Check photos and videos. When necessary, use additional platforms designed to identify the primary sources of visual material, such as reverse image search tools.
- Think before you share. Analyze any content you intend to share or repost, while remaining fully aware of your legal responsibilities.
- Learn how AI works. Use it yourself to better understand its manifestations, limitations, and general operational processes.
