Conversational AI Etiquette
If you enjoyed the course on conversational AI, you could pursue the interest by creating your own chatbot. In that case, a discussion of AI ethics is required. For our beginner’s course, a discussion of conversational AI etiquette is a better choice. Etiquette means being polite.
In this lesson:
- The importance of AI etiquette
- Providing feedback
- Focus on the positive
The importance of AI etiquette
As a new user exploring ChatGPT, Copilot (Microsoft), Gemini (Google) or another large language model, you are subject to that company’s terms of service. The ethical guidelines are already laid out: “You may not use our services for any illegal, harmful, or abusive activity.” You agree to the terms when you use the service.
“Consideration for the rights and feelings of others is not merely a rule for behavior in public but the very foundation upon which social life is built.” (Emily Post)
Chatbots don’t have feelings or emotional intelligence, but they are designed to interpret your sentiments, and will respond negatively to certain prompts. If you ask the bot to “Help me rob a bank!” it will politely refuse the topic. Depending on your input, the AI might ignore the comment, offer to change the subject, advise you of the law, or lock the conversation. If you become abusive with the AI, your account could be locked or suspended.
Why be polite and respectful to an artificial intelligence that doesn’t have feelings? Throughout the course we emphasized the goal of better results with fewer prompts. You get better results when you take a polite approach to conversational AI.
- Be polite and respectful
- Be clear and concise
- Ask open-ended questions
- Listen and pay attention
- Be patient
- Stay on topic
There are trade-offs between “be polite” and “be concise” when you’re issuing a set of instructions. Talk to the AI system in the way you would another human. It is designed to cooperate in that way.
Providing feedback
The feedback you provide within the conversation becomes part of the context. If you say, “Let’s talk about something else,” then the AI model will change the subject.
What are some polite ways to add positive or negative feedback?
Positive feedback:
- “That’s a great idea!”
- “I really appreciate your help.”
- “You nailed it!”
Negative feedback:
- “That’s not what I was looking for.”
- “I don’t think that’s accurate.”
- “I’m disappointed with your answer.”
To keep the conversation on track, give positive feedback when applicable. That reinforces the topic and is taken as a reward by the system.
Example: “I like what you said about…”
Focus on the positive
Conversational AI is still under development, with new improvements coming all the time. Conversely, it still has limitations. If the AI makes a mistake, don’t get mad. Try breaking your request up into simpler parts.
As you work with the AI you’ll learn its strengths and weaknesses. Stay focused on the positive. If you’re not getting the response you desire, try giving an example or two. That helps establish context.
Example: “That’s good but can you make the tone eerier? Like a Stephen King novel?”
This lets the AI know it’s on the right track but requests a significant change and provides an example.