prompt engineering
# Five Principles of Prompting
# What is prompt engineering?
It’s getting increasingly impossible to keep up with AI:
- published papers on arXiv growing exponentially
- open source AI projects (stable diffusion, AgentGPT) are the fastest growing projects on GitHub
- Midjourney discord server has 15M members
It is becoming increasingly challenging to stay updated with the rapid advancements in the field of artificial intelligence (AI). The number of published papers on arXiv, a popular platform for scientific research, is growing exponentially, making it difficult to keep track of the latest developments. Additionally, open-source AI projects such as stable diffusion and AgentGPT are among the fastest-growing projects on GitHub, indicating the widespread interest and active participation in AI research and development. Furthermore, communities like Midjourney discord serve as hubs for AI enthusiasts, fostering collaboration, knowledge sharing, and further accelerating the pace of AI innovation. As AI continues to evolve at an unprecedented rate, it is crucial for individuals and organizations to actively engage with these resources to stay informed and contribute to the advancement of this transformative technology.
AI results are non-deterministic
The doomer narrative that AI will take all our jobs is likely overblown, the productivity boost from working with AI is real
It’s a new form of programming, where the skill is not in writing algorithms or training your own AI models, but “curating prompts to make the meta learner get the task it’s supposed to be doing”
Things to consider in prompting: Direction, Format, Examples, Evaluation, Division
Prompt engineering is the process of discovering prompts which reliably yield useful or desired results.
Five principles of prompting:
Giving direction: Describe what you’re imagining, to get output that matches your vision
Specifying format: Define your required response format, to minimize time spent parsing errors
Providing examples: Integrate examples in your prompts, and improve the reliability of your prompt
Evaluating quality: Identify errors to iterate and improve on the reliability of your responses
Dividing labor: User the right model and the right prompt for the right job, then chain them together for sophisticated tasks