prompt engineering
The practice of phrasing queries to an LLM in a particular way to get good results, arising from the models' sensitivity and brittleness to wording.
As defined in ms_280
Other definitions
The practice and discipline of crafting input prompts to language models in ways that elicit better, more accurate, or more useful outputs; an emerging skillset as language models become more powerful
How ChatGPT changed everything: Thoughts from the frontline of the AI/ML revolution
Where it comes up · 27
- Generative AI in a Nutshell - how to survive and thrive in the age of AI9 claims
- OpenAI CEO Sam Altman | AI for the Next Era5 claims
- You need to learn AI in 2024! (And here is your roadmap)4 claims
- What the Heck is Graph Engineering?3 claims
- Tyler Cowen on how AI will reorder economies, schools, and spirituality2 claims
- Zvi Mowshowitz on AI and the Dial of Progress2 claims
- Aravind Srinivas - Building An Answer Engine - [Invest Like the Best, EP.363]2 claims
- Jeremiah Lowin: Explaining the New AI Paradigm - [Invest Like the Best, EP.307]2 claims
and 19 more
Most important claims · 10 of 51Search all claims for “prompt engineering” →
- One of Mitchell's biggest fears is that humans will trust AI systems with tasks they are not capable or robust enough to handle, as shown in Harvard Business School research where consultants using GPT-4 performed worse on quantitative reasoning tasks than those not using AI, because they over-relied on the system and made more mistakes.The Future of Artificial Intelligence
- The reason AI systems are effective is not the algorithms (gradient descent and transformers) but the data: hundreds of thousands of humans who put text and images on the web, plus reinforcement learning from human feedback and prompt engineering, supply the human exploratory and creative intelligence; tech companies then credit the algorithms much like the 'Stone Soup' folktale, in which the magic stone gets credit for soup actually made from villagers' contributions.308 | Alison Gopnik on Children, AI, and Modes of Thinking
- The idea that someone can become a great vibe coder or prompt engineer without first becoming a competent programmer is a pipe dream, because the capacity to be a good editor is the reward you earn from being a good doer—editors who can't write can't give meaningful pointers.Dhh David Heinemeier Hansson
- Prompt engineering—the skill of communicating effectively with AI—is as essential in the age of AI as reading and writing were in previous eras.Generative AI in a Nutshell - how to survive and thrive in the age of AI
- You can make an AI effectively smarter through prompt engineering—framing questions so the system imitates a smarter interaction—just as a human gives stupid answers to stupid questions; most users give up too early when the system underperforms, when more skilled prompting would have produced far better answers.Zvi Mowshowitz on AI and the Dial of Progress
- Prompt engineering, which emerged in 2023-2024, represented a recognition that work processes were shifting to deputizing AI chatbots, requiring optimization of prompts to get the best results from AI systems.What the Heck is Graph Engineering?
- The phenomenon of 'prompt engineering'—where simply adding 'Let's think through this step by step' to a prompt produces better answers from GPT-3—is unexplained and counterintuitive, as there is no logical reason why such a request should improve the model's output.This is not the AI we were promised | The Royal Society
- Genuine intelligence—the ability to think, apply logic and reason, and figure out novel things—is distinct from memory, stored knowledge, and processing speed; AI systems can imitate understanding by having many close facsimiles in their training data without actually understanding the way someone who genuinely grasped the thing would.Zvi Mowshowitz on AI and the Dial of Progress
- The most interesting AI-generated art is not autonomous machine output but a dialectic process requiring human skill, where creators with backgrounds in photography or creative direction know the right prompts and terms to achieve desired outcomes, meaning not everybody can create good AI art and the value lies in the symbiotic relationship between human and machine.326 - Ai Information Integrity
- Prompt engineering sophistication among top LLM users is vastly higher than amateur use; leaked system prompts for Claude and ChatGPT are ~10,000 words long and extraordinarily sophisticated, while average users create 2-line prompts, suggesting massive performance gap from user skill variation.Tyler Cowen on how AI will reorder economies, schools, and spirituality