AI Tools Like ChatGPT and Bard Found to Provide Invalid Citations in Cancer Research Queries

The rapid adoption of AI-powered tools, such as ChatGPT and Bard, is reshaping how patients seek healthcare information online. These platforms, driven by large language models, offer conversational and empathetic responses to health-related queries, often surpassing traditional search engines and even physicians in perceived quality and empathy. The shift from conventional search methods to AI-driven chat interfaces is poised to disrupt the healthcare ecosystem significantly, raising urgent questions for pharmaceutical companies about how to adapt to this evolving digital landscape.

AI tools are increasingly being used by patients to access detailed health insights without consulting healthcare professionals (HCPs). Biohackers have already begun leveraging AI for interpreting personal health data like bloodwork. Unlike traditional search engines that provide ranked lists of websites, AI chat platforms deliver concise answers in a conversational style that fosters trust and engagement. However, concerns about misinformation persist. A recent analysis conducted by Brand Medicine International revealed that half of the citations provided by ChatGPT during cancer research queries were invalid, while all citations from Bard failed verification. Julia Walsh, CEO of Brand Medicine International, emphasized the risks posed by these tools due to their reliance on content from the visible web—often behind paywalls or firewalls—and their tendency toward “hallucinations,” where incorrect information is presented with confidence.

Pharmaceutical leaders are urged to respond proactively as Search Engine Optimization (SEO) gives way to Search Chat Optimization (SCO). Walsh recommends retooling educational resources for compatibility with AI platforms and creating dedicated teams within pharma organizations focused on adapting communication strategies for this new environment. Additionally, regulatory reform may be necessary to ensure locally relevant and evidence-based health advice is accessible through AI-driven searches. While challenges remain—such as ensuring transparency in sourcing information—there are opportunities for pharma companies to leverage AI tools effectively. For instance, Google’s Med-PaLM 2 has demonstrated expert-level accuracy in answering medical questions based on United States Medical Licensing Examination-style queries.

Hybrid consultations combining HCP expertise with AI capabilities could further enhance patient care by filtering misinformation during interactions. Industry experts suggest collaborative efforts involving pharmaceutical companies, patient advocacy groups, regulators, and generative AI providers like Microsoft and Google as essential steps forward. Recommendations include establishing advisory boards focused on digital health literacy and designing custom chatbots using approved evidence-based content tailored for local contexts.

As stakeholders navigate this transformative period in digital healthcare communication, agility and collaboration will be critical in addressing both the threats posed by misinformation and the opportunities offered by advanced AI technologies.

Newsflash | Powered by GeneOnline AI

Source: GO-AI-ne1

Date: June 19, 2023

Author

Mark Chiang