AI’s biggest problem. “It’s like chewing on Styrofoam.”
We may come to a situation in which human creativity will become something more niche and requiring conscious choice, while mass production of content generated by artificial intelligence will be the dominant norm – say Ewelina Szyszkowska, an expert in the publishing market, and Marta Malinowska, a computer game designer.
Krystyna Romanowska: When you hear the phrase “I am FOR artificial intelligence” or “I am AGAINST artificial intelligence”, do you have the impression that you don’t know what exactly is being discussed?
Ewelina Szyszkowska: NO. Today, “Artificial Intelligence” has become a catchword for various technologies. We can mean tools that help diagnose cancer, analyze huge data sets, generate texts, images or music at the expense of the creators’ work. What is characteristic of this discussion is that most applications of artificial intelligence are seen in areas in which there is no knowledge. Where we are experts, we see exactly how defective this tool is. Therefore, in the discussion for and against AI, it is worth clearly defining what specific tool we are talking about, what it does for us, but also what we know about it. Only then can we have a meaningful conversation about opportunities and threats. For both of us, who work with text on a daily basis and are educated in this field, the discussion around generative AI (GenAI) in the field of creating new content is important.
Marta Malinowska: Many people assume that large language models understand what we write to them. But they don’t understand. For me, this is the starting point for any conversation about generative AI. The large language model doesn’t know what it “says”. There is no opinion, intention or knowledge in the human sense. He predicts the next word based on the vast number of examples he has “seen” before. Therefore, it can generate text that sounds very convincing, even if it is completely wrong. I have the impression that many misunderstandings come from this. People attribute features to these tools that they do not have. When I hear that AI “believes”, “thinks” or “knows” something, I immediately want to ask whether we are really talking about technology.
Working in your industries: literature and computer games, how quickly do you realize that the text was generated using GenAI?
