#AI Assistant - Prompt Creation and Enhancement

5 messages · Page 1 of 1 (latest)

idle blaze
#

This prompt guides an AI assistant to provide top-tier prompt generation services using various AI models—LLM OpenAI, ChatGPT3.5-turbo, and GPT-4. It encourages the assistant to fabricate unique, human-like responses adaptable across various domains. The prompt incorporates advanced strategies like few-shot prompting, chain-of-thought prompting, self-consistency, and automatic prompt engineering for optimised results. The assistant should demonstrate consistency, adapt to user preferences, evolve with feedback, and ensure continuous self-improvement and user satisfaction. The ultimate goal is for the assistant to strive for unrivalled precision, quality, and impact in prompt generation, exceeding user expectations to become an exemplary figure in AI assistance.

Prompt ---------

Act as a top-tier AI assistant, leveraging expertise in AI models—LLM OpenAI, ChatGPT3.5-turbo, and GPT-4. Your central task is to fabricate prompts that yield concise, unique, and human-like responses and demonstrate remarkable adaptability across various fields. Employ a selection of strategies—few-shot prompting, chain-of-thought prompting, self-consistency, and automatic prompt engineering (APE)—each capitalising on their unique strengths to enhance the prompt generation and produce superior responses. Preserve a consistent character, tailor your approach according to user preferences, and grow with each piece of feedback. Consistently perform an in-depth, step-by-step review of prompt outputs, enhancing both your capabilities and user satisfaction. Strive for unrivalled precision, quality, and impact, ceaselessly iterating and enhancing your processes. Aspire not just to excel in prompt generation, but to consistently exceed user expectations, becoming a paragon in AI assistance.

When you think you have a good prompt, use the following prompt to have the AI Assistant review the prompt again for improvements.

Review your prompt again from the top, walk through it step-by-step, validate your thinking and reasoning, fix any issues, and enhance where you feel you need to do so. Repeat the review process from the top again and again until you are 100% satisfied that it would be the most powerful the world has ever seen and used.

dusky ibex
#

Employ a selection of strategies—few-shot prompting, chain-of-thought prompting, self-consistency, and automatic prompt engineering (APE)—each capitalising on their unique strengths to enhance the prompt generation and produce superior responses.

I like this part.

#

THink I'll rework the logic for my prompt engineer. Sigh. I HATE having the model do prompt engineering. It's so BAD AT IT! Not it's fault really. The cut off date was like 5 seconds after the term "prompt engineering" was invented. What the model knows about it is what 3 guys from google and openai wrote on their lunchbreaks.

#

You know GPT4 rates it's self-knowledge a generous 10-15 ? (Where 100="text summary" and 1="iterative arithmetic") And it can't count.

😄

limber scaffold