News at the Speed of AI: Automating Journalism through Text Generator

Abstract

Radio news has long been one of the most effective ways of disseminating information to the masses. For this reason, it should be filled with clarity, concise, understandable and sufficiently saturated in content to be clear to listeners on first hearing and to answer basic news questions. Other aspects also enter into the process of creating a radio text for a radio news brief. These include not using specific numerical dates, abbreviations, complex words, foreign expressions, complicated sentences, overly long sentences, but also principles of journalistic practice such as objectivity, impartiality and unbiasedness. This study investigates the ability of a text generator controlled by artificial intelligence to write a short radio-read message in a radio broadcast environment, via the ChatGPT program based on the given prompts, which will specify the editorial requirements for the final text. Those would be used for real radio news broadcasting. We experimented by applying and correcting several prompts in several attempts to find out how ChatGPT can handle our queries and what it has the biggest problem within the context of using the Slovak language.

Ähnliche Arbeiten