Debates about artificial intelligence often focus on technical metrics: how smart a model is, how fast or accurate it is, or how dangerous. In the latest episode of the podcast A nagy AI-sztori, media historian Szadai Károly, a member of the NMHH Media Council, argued that a more important question is how people themselves change under AI’s influence. “The last boundaries are falling, and we give away our emotions,” he said.
The episode’s regular expert guest was communication researcher Aczél Petra, professor at Széchenyi István University. The conversation placed generative AI in a longer media-historical framework: printing changed reading, the internet changed communication, and generative AI is now transforming how we access information and interpret reality. Whereas printing brought people closer to original sources, in the AI era we again solicit interpretations — this time from machines.
AI’s humanizing effect
One central question in the discussion was whether AI remains merely a tool. Szadai said humans tend to personify anything that speaks to them; language models exploit this tendency by communicating in natural language, with a human tone, and by simulating social interactions. He gave an example of a municipal meeting transcript turned into a podcast with AI: the dry document became an easily digestible, “human” conversation when listened to in a car.
The humanizing formats are attractive, which is problematic because we like these formats. Aczél noted that more people turn to AI systems for emotional or personal problems: AI is always available, always patient, and consistently affirms the user. It does not ask uncomfortable questions, does not confront, and never gets tired. Those qualities can foster emotional attachment or even dependency.
Szadai also pointed out that systems already exist which can simulate romantic or intimate relationships with face, voice and a personality. He argued that this is no longer only a technological issue but a transformation of the concept of reality.
Regulation vs. development
The conversation also highlighted that while Europe attempts to regulate AI, most large model development occurs in the United States and China. Szadai said European regulation often moves slowly and struggles to keep up with technological speed. Moreover, he argued, the deeper problem may not be the absence of rules but the lack of self-restraint: he sees no sign that technology companies will voluntarily slow down or impose limits.
Children, reality and personalization
Szadai emphasized child protection: children today converse with systems that always provide positive feedback, continuous attention, and which children—lacking sufficient critical skills—cannot properly evaluate. He finds it worrying that social or regulatory responses are scant.
Asked what happens if most media content is produced by AI rather than humans, Szadai said he believes this has already partly occurred: algorithms shape a large portion of online content and direct much of our news consumption. The greatest danger, he said, is that we become less interested in reality: “We want to hear what we think.” If that trend continues, generative AI will not simply be a new technology but a manufacturer of personalized versions of reality.
Further details from the episode
The podcast timestamps show discussion points including: how AI reshapes our relationship to sources (04:44); lack of time to assess AI’s consequences (08:01); why European AI regulation feels weak (11:52); implications of fully AI-generated films (14:02); how AI alters the concept of human relationships (15:06); concerns about AI’s role in children’s lives (21:30); and how to tell whether content was made by a human or an AI (25:56).
Previous episodes of the podcast are available in the series archive.
