When students have to write an essay, they no longer have to pick up a pen themselves. Teacher Florian Nuxoll demonstrates an example of what this can look like: Computer on, browser open, typed in the task in the artificial intelligence Chat-GPT, and the text appears piece by piece, just as you wish. And “you won’t be able to identify it as plagiarism,” says Nuxoll – because the artificial intelligence rewrites the text every time.
Welcome to the seventh Teachers’ Media Day, an initiative of the Bavarian newspapers and the State Center for Civic Education. You can imagine the whole thing as a kind of virtual training. In Munich, there is a discussion in a studio at the Mediaschool Bayern, while interested teachers watch the stream and submit questions. More than 1,000 of them registered this Wednesday, more than ever. The topic is also very current: the influence of artificial intelligence on media and schools.
However, as the integration of AI in education grows, concerns also arise. An emerging need is the implementation of an “AI Detector” to detect ChatGPT generated text to address the potential misuse of AI-generated content. This tool would serve as a safeguard, ensuring that the educational material produced by students is authentic and that the influence of AI is harnessed responsibly.
In fact, AI has been talked about for a long time, as a kind of distant future that will shape the way we work, live and think. But since November 30, 2022, this music has sounded very present. At that time, Chat-GPT was released to the public and amazed people with answers that sounded as natural as they were intelligent. No known program had ever responded so well to human input. Since then, the tool has also found its way into everyday school life, albeit on a small scale and sometimes through the back door. In the past, homework was copied from a neighbor, and later the Latin translation was googled online. Today a chatbot sometimes helps.
However, it would be too short-sighted to see it as an under-grinding machine. Teacher Nuxoll made this clear on Wednesday. He tells how he got inspiration for worksheets and letters to parents from Chat-GPT. For example, students could be guided in structuring a discussion. The Ministry of Culture is also convinced that AI can support learning and save time for teachers. Pilot tests are therefore being carried out at selected schools in which, for example, learning software is tested in English lessons.
In the first attempt, Chat-GPT failed the Abi, the second time the AI passed
In order for AI to have a targeted effect in schools, teachers are “the key,” as Minister Anna Stolz put it in a video message. But even experts find it difficult to keep an eye on developments over the past twelve months alone. When Bayerischer Rundfunk had Chat-GPT write the Bavarian Abi in January, the bot failed. In May, the successor version GPT-4 managed the questions quite easily. Hey-Gen now even has an AI that lets you speak in foreign languages. All you need is a video of the user with the desired sentences, the machine translates and imitates the rest. Lip sync. Nuxoll asks provocatively whether there are still foreign language subjects needed. His thesis: “Artificial intelligence will change the education system more sustainably than any other innovation since the introduction of compulsory schooling.”
But artificial intelligence can also influence everyday school life in other ways: in the form of fake news and image falsifications, as a vehicle for propaganda and hate speech. AI-generated manipulations are still in the minority compared to human ones, reports journalist Stefan Voß. But it doesn’t have to stay that way. Voß heads the verification department at the German Press Agency, so he has experience in recognizing fakes. There’s a lot of it floating around the internet at the moment – and through the smartphones of children and young people. It is often difficult to recognize what is true and what is false. Voß advises, even though it sounds unsatisfactory: “Relax.” Verification takes time. With a calm look, you often noticed details that didn’t fit the message. The picture shows a burning hotel, but the people are sitting quietly on the beach next to it? Nobody takes out their cell phone to record the event? Clearly a fake. “Logical thinking,” says Voß, “is the be-all and end-all.”
Seen this way, artificial intelligence can be viewed as dangerous, helpful, or both at the same time. In any case, they are unlikely to disappear back into the futuristic department. Voß therefore reminds us of a piece of journalism wisdom: “You can only report on it if you deal with it.” And Nuxoll makes an appeal to his colleagues: “If you don’t retire in the next few weeks, then you have to learn how to do it.”