Artificial intelligence (AI) is rapidly integrating into news and content, prompting a necessary reflection on its implications for democratic societies, which rely on trustworthy and diverse media sources. A new report, Artificial intelligence and media policy: Plurality from the meat grinder, from Professor Dr. Rupprecht Podszun, Heinrich Heine University, and Ruth Meyer, Director of the Saarland State Media Authority (LMS), delves into the potential risks AI poses as it is applied in the media industry as well as the need for regulation to govern its usage.

Guardrails for AI

Podszun and Meyer identify three areas needing guidelines and policies to uphold democracies:

1. Trust in information is about ensuring that the information people receive is accurate and reliable, whether it’s news or other content. Laws like the Saarland Media Act stress the importance of journalists being careful and accurate when reporting. However, when artificial intelligence involves creating content, it can be hard to know if the information is trustworthy because the processes behind AI are often hidden. This lack of transparency can make people doubt the reliability of AI-generated content.
2. Public discourse refers to the conversations and discussions in society, especially around important issues. With the rise of AI-powered recommendation systems, people often get information that aligns with their beliefs and interests. This can create “bubbles” where people only hear opinions like their own, leading to societal divisions. The idea of the public sphere, where people from different backgrounds come together to discuss and solve problems, is important for democracy. However, social media platforms, which play a big role in public discourse today, can make it harder for diverse opinions to be heard.
3. Plurality is under pressure despite the many different media sources available today. The reality is that companies like Alphabet (Google’s parent company) and Meta (formerly Facebook) have too much control over the information people see. This concentration of power can limit the variety of perspectives and ideas people are exposed to, especially when AI algorithms are used to personalize content delivery.

A regulatory approach for AI

Regulation will be needed to address concerns and provide guidance for the most constructive development of the AI industry and its applications. This could include laws to prevent monopolistic companies from having too much control over information. Further, there is a need for transparency about how AI is used in creating content. It’s also important to ensure that the data used to train AI systems is diverse and representative of different viewpoints.

The authors provide recommendations for regulatory approaches to ensure diverse and trustworthy access to information. They suggest that:

• Preventing monopolistic concentration is crucial, given the dominance of big tech companies in both data-driven business models and AI control. Media concentration laws should counteract this trend, promoting diverse data pools and open technology. 
• Ensuring transparency and responsibility are fundamental. Trust in AI-driven media necessitates transparency regarding its usage, training data, and information sources. 
• Identifying prohibitions is necessary to enforce accountability. If AI crosses ethical boundaries, explicit bans with sanctions are imperative. 
• Addressing the training data problem is vital. Guaranteeing open and diverse data selection mitigates distortions in AI-generated content. Embracing adaptable AIs capable of correcting errors ensures ongoing development and integrity.

While AI offers unprecedented opportunities for media innovation, its unchecked proliferation poses significant risks to democratic principles. Effective regulation is imperative to harness the potential of AI in media while safeguarding pluralism, transparency, and trust in information dissemination. Only through collaborative efforts between policymakers, media stakeholders, and technologists can the transformative potential of AI be harnessed responsibly in the service of democratic societies.