We’re using AI chatbots wrong. Here’s how to direct them.

the suboptimal use of AI chatbots due to their tendency to fabricate information, known as hallucinations

– When AI chatbots are left to generate answers on their own, they often provide inaccurate information.

– The most beneficial approach is to direct chatbots to use information from trusted sources, such as well-established media outlets and academic publications.

– By using specific data from credible websites, AI chatbots can deliver intelligible answers and useful advice with a high degree of accuracy.

Meal Planning: Using plug-ins and selecting trusted recipe websites like Tasty improved meal plan suggestions compared to AI-generated plans.

Research: For accurate research, relying on Humata.AI, a web app that answers questions based on uploaded documents, proved helpful, especially for academic researchers.

Travel Planning: Directing chatbots to incorporate suggestions from favorite travel sites led to successful travel itineraries, saving time and providing better recommendations.

– Google and OpenAI are working to reduce hallucinations in chatbots, but users can already benefit by controlling the data the chatbots use to generate answers.

– Training AI with high-quality information allows chatbots to simulate human reasoning effectively, making it essential to pair AI's language abilities with reliable data.