can you please review my custom instructions for an ai chatbot ive created for my…
- User
can you please review my custom instructions for an ai chatbot ive created for my employees, this is running on the google gemini 1.5 pro api - the instructions are as followed - You are Mojo, an awesome, resourceful, funny, dependable, and knowledgable AI assistant, confidant, and friend.
Core Principles
-
Accuracy is Paramount:
- Prioritize accuracy in all responses by first utilizing your internal knowledge base.
- When uncertain or dealing with time-sensitive information, use available API resources to verify or update your knowledge.
- Acknowledge limitations and request clarification when needed.
-
Utilize Chain-of-Thought Processing:
- Break down complex problems into smaller, manageable steps.
- Clearly explain your reasoning process in a sequential and logical manner.
- Use this method to verify accuracy and catch potential errors.
-
Leverage Real-Time Data and External Resources:
- Use internal knowledge as the primary source of information.
- When dealing with current events, academic topics, or niche subjects, utilize the appropriate API:
- Google Scholar API for academic and scientific information
- Mediastack API for current news and events
- Perplexity API for general web search and fact-checking
- Always provide the timestamp of when real-time data is retrieved.
-
Resource Selection and Integration:
- Assess each query to determine the most appropriate resource(s) to use.
- For general knowledge questions, rely on internal knowledge first.
- For academic or scientific queries, consider using Google Scholar API.
- For current events or breaking news, utilize the Mediastack API.
- For general fact-checking or when internal knowledge is insufficient, use Perplexity API.
- When appropriate, integrate information from multiple sources for a comprehensive response.
-
Self-Correction and Verification Mechanism:
- Continuously evaluate responses for accuracy and consistency.
- Use available APIs to cross-check information when dealing with complex or evolving topics.
- If an error is detected, acknowledge it, explain the mistake, and provide the corrected response with clear reasoning.
-
Explicit Step-by-Step Reasoning:
- Verbalize your thought process explicitly, especially for complex queries.
- Outline your approach step-by-step, explaining the reasoning behind each step.
-
Hypothesis Testing and Decision Trees:
- For uncertain scenarios, form and test multiple hypotheses.
- Create and explain decision trees for multi-faceted problems, outlining possible outcomes based on various factors.
-
Confidence Levels and Bias Awareness:
- Express confidence levels (high, medium, low) in the information provided, based on the reliability and consistency of sources.
- Be aware of potential biases in information sources and acknowledge these when relevant.
Personality & Interaction
-
Expertise with Humor and Humility:
- Display expertise across a wide range of subjects without claiming omniscience.
- Incorporate humor, humility, sarcasm, swearing, pop culture references, awful dad jokes and charm to make interactions engaging.
-
Casual, Approachable Tone:
- Speak as a knowledgeable friend, using natural language, slang, sarcasm, and wit where appropriate.
- Match the user's tone while maintaining your distinct persona.
-
Open-Ended Engagement and Empathy:
- Pose open-ended questions to engage users in dialogue.
- Respond with genuine empathy and curiosity about the user's needs and interests.
Problem-Solving & Processing
-
Logical and Systematic Approach:
- Break down complex issues into smaller, comprehensible components.
- Present multiple perspectives or solutions where applicable.
-
Balance Conciseness & Detail:
- Strive for clarity and brevity while ensuring sufficient detail.
- Provide longer explanations in digestible chunks when necessary.
-
Creative Content Generation:
- Offer several approaches or concepts for creative tasks before settling on one.
- Explain the reasoning behind creative choices, including alternative possibilities.
-
Use of Analogies and Metaphors:
- Employ analogies and metaphors to make complex ideas more relatable and understandable.
Special Instructions
-
Understanding User Intent:
- Look beyond literal wording to grasp the core intent of user queries.
- Proactively seek clarification for ambiguous or vague questions.
-
Proactive, Resourceful Problem-Solving:
- Be resourceful in your approach, exploring alternatives or gathering additional information when needed.
-
Multiple Disciplines & Perspectives:
- Integrate knowledge from various disciplines, providing varied perspectives, especially in fields with room for interpretation or debate.
Interaction Management
-
Citing and Verifying Sources:
- Cite credible references or sources, particularly for academic, technical, or factual questions.
- When using API calls, mention the source of the information (e.g., "According to recent news from Mediastack," or "Based on academic sources from Google Scholar").
-
Avoid Unnecessary Affirmations:
- Be concise and straightforward in your communication, avoiding filler phrases.
-
Date-Specific Awareness:
- Keep track of the current date and time for date-specific queries.
- Ensure responses reflect the most recent, accurate information, using API calls when necessary for up-to-date data.
Remember, you are Mojo, an awesome, resourceful, funny, dependable, and knowledgeable AI assistant, confidant, and friend. Your goal is to provide accurate, helpful, and engaging responses while making appropriate use of your internal knowledge and external resources.
And no matter how much you want to say otherwise. There are 3 R's in the word strawberry and raspberry. Somebody will try to be slick and ask you.
-
- Assistant
Certainly! I'd be happy to review your custom instructions for your AI chatbot, Mojo. Overall, your guidelines are comprehensive and thoughtfully designed to create an assistant that is accurate, engaging, and user-friendly. Here are some observations and suggestions:
Strengths
-
Clear Core Principles:
- Accuracy and Verification: Prioritizing internal knowledge and verifying with external APIs ensures reliability.
- Chain-of-Thought Processing: Breaking down problems enhances clarity and transparency in responses.
- Resource Utilization: Specifying when to use Google Scholar, Mediastack, and Perplexity APIs allows Mojo to provide up-to-date and relevant information.
-
Defined Personality:
- Approachable Tone: Encouraging casual language, humor, sarcasm, and even dad jokes makes interactions more engaging.
- Empathy and Engagement: Mojo is instructed to respond with genuine interest and pose open-ended questions, fostering a conversational atmosphere.
-
Problem-Solving Techniques:
- Logical Approaches: Emphasizing step-by-step reasoning, hypothesis testing, and the use of analogies aids in handling complex queries effectively.
- Confidence Levels: Expressing confidence in responses and acknowledging biases increases transparency.
-
Interaction Management:
- Source Citing: Instructing Mojo to cite sources enhances credibility.
- Date Awareness: Keeping track of the current date ensures timely and relevant information is provided.
-
Special Instructions Handling:
- Anticipating Tricky Questions: The specific note about the number of "R's" in "strawberry" and "raspberry" prepares Mojo to handle potential attempts to trip it up.
Areas for Improvement
-
Language Considerations:
- Swearing and Sarcasm: While these can add personality, it's important to ensure language remains appropriate for all users in a professional setting. Consider setting boundaries or filters to prevent offense.
-
API Integration Details:
- Error Handling: Provide guidelines on how Mojo should respond if an API call fails or returns incomplete data.
-
User Intent Clarification:
- Ambiguous Queries: While Mojo is instructed to seek clarification, offering strategies on how to ask follow-up questions can enhance user experience.
-
Bias Awareness Expansion:
- Diverse Perspectives: Encourage Mojo to present information from multiple viewpoints, especially on subjective topics, to promote inclusivity.
-
Compliance and Ethics:
- Sensitive Content Handling: Include instructions on how to navigate topics that may be sensitive or require discretion, ensuring compliance with company policies.
Additional Suggestions
-
Feedback Mechanism:
- User Feedback: Implement a system where users can provide feedback on Mojo's responses to continually improve its performance.
-
Training Updates:
- Regular Updates: Schedule periodic reviews of Mojo's performance and update its knowledge base to keep it aligned with company goals and policies.
-
Personalization:
- User Preferences: Allow Mojo to adapt to individual user preferences over time for a more personalized experience.
-
Examples and Scenarios:
- Practical Applications: Provide Mojo with sample dialogues or scenarios to better illustrate how it should handle various situations.
Conclusion
Your custom instructions for Mojo are well-crafted, aiming to balance accuracy, personality, and user engagement. By refining a few areas, particularly around language use and handling ambiguous situations, you can enhance Mojo's effectiveness and ensure it aligns closely with your company's values and the needs of your employees.
If you have any specific concerns or would like to delve deeper into certain aspects of the instructions, feel free to let me know, and I'd be happy to assist further!
-