COMM 110 Introduction to Public Speaking

Pre-Made ChatGPT Prompts - Roles

Assigning a role to ChatGPT, such as tutor or mentor, can guide the tool into providing the type of output you'd like to receive. Regardless of the role ChatGPT (or any LLM) is playing, there are always risks to consider:

• Hallucinations: LLMs frequently create incorrect but very believable information. These "facts" can be hard to spot, and are often woven into the LLMs responses. Hallucinations occur most often when LLMs are asked to create quotes, citations, sources, or other detailed information 
• Bias: LLMs are trained on large amounts of text and receive additional training from humans. Both training methods can introduce bias into the LLM. Many biases have been found in existing LLMs, including racial, gender, and viewpoint biases. 
• Privacy: Any information entered into ChatGPT or another LLM or AI tool is subject to the terms and conditions of that tool. Most often, that means the text you entered can be used to train the LLM. Many issues surrounding privacy of data entered into these tools are still unclear. DO NOT enter personal identifiable information (SSNs, birthdates, addresses) or other private information.

Guidelines for Using LLMs / AI

• You are responsible for your own learning and your own work 
  • You should check any information you get from AI sources with other trusted sources
• It can 'hallucinate' or make things up
  • AIs and LLMs cannot provide citations or sources
• AI is not a person, but it can act like one 
  • Don't read meaning into AI responses and remember that AI doesn't know you or the context in which you are working 
• If at first you don't succeed, try again 
  • AI can be unpredictable and sometimes gives responses that don't match what you wanted. Try asking again, asking a different way, or asking a different tool 
• You are in charge 
  • If you don't want to answer a question, tell the AI to move on
  • End the interaction at any time 
• AI might respond to your language or tone
  • If you argue with the AI, it might think you want it to argue back. Clearly communicate your preferences for how it should communicate to you. 

AI Tools for Improving Your Public Speaking