A recent study undertaken by researchers from the University of Kansas Life Span Institute has revealed a surprising trend among parents seeking health information for their children. The study found that, in situations where the author of health advice was unknown, parents were more inclined to trust AI-generated content over that provided by healthcare professionals. This development has sparked discussions about the implications of relying on artificial intelligence (AI) for crucial health decisions, highlighting potential risks associated with the accuracy of AI-generated information.
The study, which is detailed in the Journal of Pediatric Psychology, involved 116 parents aged between 18 and 65. These participants were presented with health-related texts on topics such as infant sleep training and nutrition. The texts were sourced from both AI, specifically ChatGPT, and healthcare professionals. Importantly, the participants were unaware of the source of the information they were reviewing.
Lead researcher and University of Kansas doctoral student in clinical child psychology, Calissa Leslie-Miller, explained the motivation behind the study. "When we began this research, it was right after ChatGPT first launched—we had concerns about how parents would use this new, easy method to gather health information for their children," she noted. Given the increasing trend of parents turning to the internet for health advice, the study aimed to examine the dynamics of this information-seeking behaviour and identify areas of concern.
The study's findings were unexpected. Parents rated AI-generated texts as more trustworthy, accurate, and reliable compared to expert-generated content, even in cases where they couldn't discern the source. Leslie-Miller expressed surprise at these results, especially considering that the study was conducted at an early stage of ChatGPT’s public availability.
One of the primary concerns arising from this preference for AI-generated content is the potential for "hallucinations" in AI responses—instances where AI provides incorrect information due to insufficient context. Leslie-Miller emphasized the significance of addressing such issues, particularly in the context of child health, where erroneous advice could have severe consequences.
Moreover, the findings indicate that AI, when prompt-engineered, can influence behavioural intentions regarding medication, sleep, and diet decisions. Such revelations underline the necessity for expert oversight when using AI as a source of health information.
Leslie-Miller and her team suggest that while AI has the potential to be an advantageous tool for parents seeking health information, it is vital for parents to corroborate AI-sourced information with healthcare professionals to ensure its validity. While AI can generate vast amounts of information efficiently, it lacks the nuanced judgment and expertise that healthcare professionals provide.
The study has broader implications beyond paediatric health, as the potential for AI-generated misinformation extends to various fields. Leslie-Miller advises caution when consulting AI for advice, recommending users rely on information from platforms that incorporate expert verification.
As AI continues to embed itself into various facets of daily life, understanding its limitations and the importance of expert collaboration will be crucial. This study serves as a significant exploration into the societal impacts of AI and emphasizes the ongoing need for vigilance and critical evaluation in the age of digital information.
Source: Noah Wire Services