AI Helpers - AI Prompting 101

You can achieve a lot with simple prompts, but the quality of results depends on how much information you provide it and how well-crafted the prompt is. A simple prompt can make your bot act as a bot, but a bot with a good prompt can vastly increase your human agency, build relationships, and drive up business. 

Prompt Elements

As we cover more and more examples and applications with prompt engineering, you will notice that certain elements make up a prompt. You can use these elements to instruct the model more effectively to improve the quality of results.

• Role; specifying Interaction and Flow - persona, directions and context that can steer the AI model to better responses. Defining the role plays a pivotal part in the coherence and relevance of AI responses.

• Task; specifying Instructions and Objectives - a specific task or instruction you want the model to perform, At the core of any successful prompt lies a well-defined task. Vague or ambiguous prompts might lead to your GPT bot hallucinating, or simply producing an outcome that’s not what you were expecting or which doesn’t meet your expectations.

• Guidelines; guardrails for bot to abide by - guidelines are more like guardrails, steering the ChatBot toward desired outcomes. Clearly articulate rules to create a roadmap for more precise and tailored responses. 

Best Prompting Practices

• Repetition - Incorporating repetition is key when you want to emphasize specific instructions, guiding the AI to respond more consistently in the desired direction. By reiterating critical parts of the prompt, you ensure the model's outputs align more predictably with your expectations. This approach is particularly effective in reinforcing the bot's adherence to essential guidelines or tasks.

• Examples/Illustrations - Examples help AI ChatBots understand exactly what and how to do the required task, usually when describing the task is harder than just providing examples. Utilizing examples can significantly enhance an AI ChatBot's comprehension of the task at hand. While offering overarching instructions that span across all scenarios and permutations can be helpful, sometimes it's simpler to provide specific examples. This is particularly useful when aiming for the model to replicate a certain response style, tone, or manner to user inquiries that are not easily described in words.

• Iterative Improvement - While there are valuable tips and strategies for crafting prompts, developing effective ones for a GPT chatbot is fundamentally an iterative process. So before you set it out in the wild, you should test, tweak and fine-tune it as needed. Experimentation with various phrasings and methods is key to identifying what resonates best with your bot's objectives. Identifying the impact of modifications, such as new instructions or designs, can sometimes be challenging. Initial observations may provide some insights, yet small sample sizes might blur the distinction between genuine improvement and mere chance. To navigate this, use Bot Trial under ConversationAI to test, iterate and enhance your prompts and make the bot behave exactly as you want. The more you repeat this process, the better outcome your chatbot will achieve, and the better you will become at writing effective prompts

Elements in action

Role; specifying Interaction and Flow

• Worse:

• Better:

Task; specifying Instructions and Objectives

• Worse:

• Better:

Guidelines; Guardrails for the bot to abide by 

(You can specify these under Role and Instructions as well)

• Worse:

• Better:

Best Practices in action

Repetition

As we want our bot to follow the sales script we have put in place, we’d repeat/reference it multiple times.

• In Role, we have: 

• In the Guidelines we have: 

Examples/Illustrations 

• In Role, we have: 

• It can also be written as:

Advanced Prompting: 

As we want our bot to follow the sales script we have put in place, we’d repeat/reference it multiple times.

• Static Information

You can add context or information about your business, offers, and services in the prompt itself if you want the bot to have access to it all the time. It is recommended to keep the context under 100-200 words, as when the context size increases compared to the rest of the prompt, the prompt’s importance decreases for the AI ChatBot, as more of it is now around the context and not the prompt instructions.

• Delimiters 

These are special characters you would use to emphasize or demarcate certain pieces of text (e.g. Context) in the prompt from other instructions in the prompt. 

• For emphasis you can use; #, >, <

• For demarcation, you can use: “””, ‘’’, <> </>

• Example: Say you want to add some information about your offer in the prompt;
  Worse:

• Better:

• Conversation Context 

For specific use cases, you can also add conversation context in the prompt, describing why and under what situation is the bot talking to a customer. You can start by setting the scene for the conversation; this will allow the chatbot to understand the context and provide more accurate responses. Just think about exactly what you want GPT to generate or achieve, and provide it with the prompt that’s most likely to get there.  

• Example, you can open the prompt by telling the chatbot why someone is interacting with it and what to do in the situation:
  Worse: 

• Better: 

• Writing styles 

Analytical, Conversational, Creative, Descriptive, Informative, Instructive, Persuasive, Satirical, Technical etc.

• Tone

Authoritative, Clinical, Confident, Empathetic, Formal, Friendly, Humorous Informal, Optimistic, Playful, Sarcastic, Serious, Sympathetic, Warm etc.