Artificial intelligence (AI) is rapidly transforming the media industry, and specialized AI tools like OpenAI’s Custom GPTs are opening up new possibilities. Our white paper explores how Custom GPTs can make journalistic tasks more efficient, faster, and more accurate—from content creation and trend analysis to political reporting. It consolidates our research and project experiences, showcasing how AI can be effectively implemented in journalism while addressing key challenges. Core topics include an introduction to OpenAI's Custom GPTs and their impact on the media world, insights into applying and developing Custom GPTs for media-specific tasks, and five practical use cases.
This white paper was created following the release of OpenAI’s Custom GPTs in late 2023 as part of the Media Lab Bayern's Future of News Fellowship.
In today’s media landscape, increasingly shaped by artificial intelligence (AI), Generative Pre-trained Transformers, or GPTs, are taking center stage. This technology, with its ability to autonomously generate content, raises important questions for journalists and media professionals: How can these advancements be responsibly integrated into daily journalism, and what impact will they have on the production and distribution of news and information?
The “Journalism, Media, and Technology Trends and Predictions 2024” report from the Reuters Institute continues to emphasize AI’s disruptive power, particularly its potential to fundamentally transform the media industry in 2024. According to the report, media companies should strive to leverage AI’s capabilities this year while also addressing its risks. Key risks in the media sector include misinformation, poor training data—resulting in systemic biases in AI-generated outputs—privacy concerns, and, most recently, copyright issues. Notably, in December 2023, The New York Times filed a copyright infringement lawsuit against OpenAI and Microsoft, claiming their content had been used to train AI models without permission. OpenAI argues that using publicly available data, including Times articles, falls under Fair Use and requires no licensing. The company views Fair Use in this context as the legitimate use of publicly accessible materials online for AI training, supported by international precedents and regulations. Nonetheless, media professionals remain aware of the immense potential of ChatGPT and other AI tools.
The goal of this white paper is to provide an understanding of the applications and challenges of OpenAI’s Custom GPTs in the media sector. It aims to bridge the gap between technical feasibility and the real-world applications that GPT technology can offer for the media industry.
Applications in the Media Industry: Custom GPTs can be utilized for a wide range of tasks, from content creation and automation to data organization and categorization. For example, they could help produce news content more efficiently or analyze complex data for journalistic research.
Benefits for Journalists and Publishers: Automating routine tasks could give journalists more time for in-depth research and high-quality journalism. Publishers could benefit from more efficient content creation (including for new audiences) and management, leading to increased productivity.
This document provides a detailed introduction to the fundamentals of OpenAI’s GPT technology, followed by insights into the experience and development of custom applications using Custom GPTs.
By summarizing practical experiences with Custom GPTs, this white paper serves as a guide for media professionals interested in exploring AI’s possibilities in their work and understanding its potential for the future of media and their own roles.
Several specific applications are highlighted in this paper. These examples are intended to inspire and illustrate potential uses. The five selected Custom GPTs are as follows:
FeedbackGPT
Brief Description: The primary purpose of this specialized Custom GPT for constructive journalism is to provide precise and highly critical analyses of journalistic content to enhance journalism quality. This is achieved through a well-structured analysis process, including clear steps from text review to providing concrete improvement suggestions based on relevant reference documents. Transparency, efficiency, and constructive feedback are central to its approach.
LokaljournalismusGPT
Brief Description: The Lokaljournalismus GPT aims to enhance exemplary local reporting in Augsburg by analyzing demographic data and accessing up-to-date local information. It provides editorial teams with analyses and feedback to better tailor content to the interests and needs of the local community. This is achieved by carefully analyzing and referencing relevant demographic documents, providing concrete, fact-based feedback, and, if necessary, accessing current information via specific interfaces with the Augsburger Allgemeine and the city of Augsburg.
TrendGPT
Brief Description: TrendGPT specializes in supporting lifestyle journalists by identifying new trend topics, providing concise, precise insights into current trends with sourced references from real-time data or supplied documents. TrendGPT offers a wealth of up-to-date information through access to top search trends on Google in Germany, along with a curated collection of documents on various trend forecasts to ensure relevance and accuracy in each response.
BundestagsdatenGPT
Brief Description: The primary function of BundestagsdatenGPT is to support journalists with precise and efficient access to and processing of Bundestag data. Its goal is to answer inquiries regarding parliamentary data while incorporating a critical journalistic perspective. Access to the Bundestag is facilitated through interfaces providing information such as printed documents, agendas, plenary minutes, and access to “Heute im Bundestag” (hib) updates.
USAPolitikGPT
Brief Description: The main purpose of USAPolitikGPT is to provide fact-based, well-researched information exclusively on U.S. politics. It delivers clear, concise, and impartial answers supported by citations. This is achieved through web research and the use of the official GovInfo API, which provides up-to-date government documents and reports as references, including economic reports, congressional documents, and legislative summaries.
This white paper aims to present and explain initial experiences with the Custom GPT Builder in a media context. We are, of course, aware of the challenges associated with using OpenAI services, such as copyright, data privacy, and biases in training data. Nevertheless, we hope to showcase how we’ve utilized OpenAI’s Custom GPTs and inspire other media professionals to experiment with Custom GPTs and identify their own use cases. Work on and with Custom GPTs began in fall 2023, directly following OpenAI’s release, and continues to this day.
The full white paper is available for download here in German:
Comments