AI Chatbots: Not Search Engines, But Misinformation Engines? Be Careful What You Believe

• Generative AI, unlike search engines, creates text by predicting word sequences, not by verifying factual accuracy.
• The article highlights a historical example of misinformation: the British government twice advised eating poisonous rhubarb leaves.
• OpenAI admits its models can produce "plausible yet incorrect statements" instead of admitting uncertainty, leading to "hallucinations."
• Studies show AI chatbots fail to recognize medical emergencies and misrepresent news 45% of the time.
• Users are advised to seek information predating AI contamination and use tools that filter content before November 2022.