The Rise of AI Overviews and Their Impact on Public Perception
Philip Morris International (PMI) is a company with a complex and controversial history. Known for its role in the global tobacco industry, PMI has long been criticized for its aggressive marketing of addictive, cancer-causing cigarettes that have contributed to smoking-related deaths and diseases worldwide. However, when a colleague recently searched for information about the company on Google, they were surprised by what they found.
Google’s AI overview feature, which provides a “snapshot of key information about a topic or question” at the top of search results, generated a summary that was notably positive. According to the AI-generated content, PMI was described as a “leading international tobacco company working to transition from cigarettes to smoke-free products.” It also highlighted the company’s efforts in reducing post-consumer waste and improving sustainability within its supply chain.
This upbeat portrayal drew criticism from experts like University of Sydney public health professor Becky Freeman, who described it as “essentially a regurgitation of Philip Morris International’s PR materials.” She pointed out that the AI overview omitted any mention of negative press surrounding the company, including its history of misleading the public about the health risks of smoking.
A Pattern of Promotional Content
Freeman is not alone in her concerns. The ABC conducted similar searches for other companies with complex histories, such as British American Tobacco and James Hardie. In some cases, the AI overviews relied heavily on information from the companies’ own websites, resulting in a promotional tone that overlooked their problematic legacies.
For example, an AI overview for James Hardie, a former Australian manufacturer of asbestos products, described the company as a “global leader” and “pioneered asbestos-free fibre cement.” However, it failed to mention the ongoing health impacts of asbestos, including thousands of annual deaths in Australia and the country’s high mesothelioma rates.
This pattern raises questions about how AI overviews are shaping public perception. Experts argue that when AI summaries draw primarily from a company’s own website, they risk presenting a one-sided view that omits critical context.
The Business of Optimizing AI Search Results
Google does not allow companies to pay directly to influence how their brand appears in AI overviews. However, the company acknowledges that these summaries can sometimes provide inaccurate content. Despite this, businesses are increasingly focusing on optimizing their websites to ensure they are featured in AI search results.
This practice, known as generative engine optimisation (GEO), answer engine optimisation (AEO), or AI search engine optimisation (AI SEO), involves structuring online content in a way that makes it more likely to be selected by AI systems. While none of the companies contacted by the ABC confirmed using these strategies, the pressure to appear in AI results is growing.
The Dynamic Nature of AI Search Results
One challenge with AI search results is their unpredictability. AI summaries can change even when the same query is used, due to factors like the wording of the question, current events, and user behavior. This means that the information presented in AI overviews may vary significantly depending on when and how a user conducts a search.
For instance, after the ABC highlighted the Philip Morris AI overview, subsequent results included a note acknowledging the company’s controversial practices. Google attributed this change to the dynamic nature of its AI overviews, which adapt based on the most relevant and timely information available.
The Influence of AI on Web Traffic and User Behavior
The rise of AI-powered search is also reshaping how users interact with the web. Eye-tracking research suggests that users engage significantly with AI overview content, while many web publishers and news outlets report declining traffic as fewer people click through links.
Some preliminary studies indicate that the way information is presented in AI overviews might even influence user attitudes. For example, data from a Melbourne-based online marketing agency showed that users referred by AI were more likely to become customers than those coming from personal recommendations.
Research and Regulation in the Age of AI
Researchers are still trying to understand the full impact of AI on information consumption. Queensland University of Technology (QUT) researchers Ashwin Nagappa and Kateryna Kasianenko are examining how search engines and AI overviews shape public perception as part of the Australian Search Experience 2.0 project.
Their findings suggest that AI overviews tend to give prominence to commercial sources, though it remains unclear whether this is a broader trend or limited to specific topics. Google maintains that it does not prioritize commercial sources but instead draws from the most reliable and helpful information available.
Despite these assurances, experts warn that the lack of clear policies or regulations around AI content presentation leaves room for bias and misinformation. As AI continues to evolve, users must remain cautious and critically evaluate the information they encounter.
