jasdooit

With generative artificial intelligence (AI) such as ChatGPT becoming a main search method, studies indicate that data is being altered during the process. AI has now taken on the position of a “gatekeeper” in deciding which media outlets are referenced.

On the 2nd, the UK Institute for Public Policy Research (IPPR) published a report addressing this concern. The findings indicated that AI systems depended significantly on a limited number of major news outlets, with ChatGPT and Google’s Gemini failing to reference the BBC at all. The report suggested that the AI news landscape might be dominated by a few technology firms and called for governmental action.

**◇New Winners and Losers**

The IPPR examined the responses of four leading AI platforms (ChatGPT, Gemini, Perplexity, and Google AI Overviews) to queries related to news. After analyzing 100 news-related questions and more than 2,500 links, it was found that the BBC—recognized as the most trusted news source in the UK—was not referenced at all in ChatGPT or Google Gemini replies. On the other hand, The Guardian appeared in 58% of ChatGPT responses and 53% of Gemini responses, making it the most frequently mentioned outlet.

The report linked this occurrence to whether AI companies had “licensing agreements” with media sources and whether news organizations had implemented “AI crawling blocks.” Indeed, The Guardian has collaborated with OpenAI, whereas the BBC opposes unauthorized data gathering by AI and restricts access. ChatGPT’s primary references mostly matched media outlets that had licensing deals with OpenAI, and each AI tool demonstrated a distinct inclination towards certain sources.

The IPPR highlighted that these practices lead to new groups of winners and losers. Over-reliance on particular media limits the variety of perspectives users are exposed to and may magnify certain opinions or agendas without users realizing it. Smaller or local media outlets, which have difficulty obtaining licensing agreements, are also excluded from AI-generated responses.

**◇Collapse in Click-Through Rates?**

The study examined the effect of widespread AI use on media revenue structures. Following the launch of Google’s “AI Overviews” function, the click-through rate (CTR) from search results to news websites decreased from 15% to 8%. As per the report, news organizations anticipate a 43% reduction in traffic driven by search engines over the next three years.

Furthermore, AI chatbots were discovered to extract content significantly more often than they generated traffic. OpenAI had a ratio of 179:1, Perplexity 369:1, and Anthropic 8,697:1. Even though these AI companies frequently scrape media content, they do not provide payment to publishers, endangering a media environment that depends on advertising and subscription income.

Similar disputes have arisen in South Korea. In April of the previous year, the Korean Association of Newspapers informed the Korea Fair Trade Commission, claiming that Naver utilized news content without authorization to create its “HyperClova” large language model (LLM). On the other hand, certain media organizations have monetized data sharing. Axel Springer, which owns Politico and Business Insider, entered into a licensing agreement with OpenAI. News Corp, which publishes The Wall Street Journal, established a five-year, $250 million content supply contract with OpenAI beginning in 2024.

**◇”Government Must Establish Systems”**

The IPPR condemned the present AI media landscape, describing it as “dominated by a limited group of technology firms that lack openness and responsibility,” and urged for proactive governmental involvement. The study suggested three approaches:

Initially, creating a “joint licensing market” to guarantee equitable payment for local and smaller media via a government-established licensing task force.

Second, implementing a “news nutrition label” system to openly indicate to users which sources and perspectives are referenced in AI-generated content.