🔗Bing Chat and Google SGE frequently linked to recent, expert-created resources, enhancing their credibility.

💬 Bard consistently included AI disclosures, promoting transparency, while ChatGPT did so less frequently.
📌 ChatGPT showed high content originality.

All models scored higher than the recommended 8th grade readability level for health information, with ChatGPT and Bard providing more complex responses. Google SGE often provided shorter, more concise answers

ChatGPT performed relatively better than others in accuracy, clarity, completeness, clinical utility, and overall rating. Bing Chat and Google SGE had lower overall scores, highlighting the areas for improvement among the models.

🔍 We evaluated how well LLMs respond to caregivers of children with cancer in finding reliable information, focusing on 4 LLM-based services available: ChatGPT (GPT-4), Google Bard (Gemini Pro), Microsoft Bing Chat, and Google SGE (Google's AI integration to search engine).

This isn’t just about tech—it’s about people. Let’s rethink how AI can truly support healthcare professionals.

Key considerations:
💻 Are AI tools functional, intuitive, and well-integrated?
👂 Are we listening to radiologists about their needs, or prioritizing high-accuracy tasks without regard to usability?
🌐 Is organizational culture ready for AI adoption?

This sparks crucial questions about how AI tools are designed, integrated, and used in real-world clinical practice. Are we truly helping providers, or adding complexity to their work? 🩺🤖