prompt
An input or instruction given to an AI system to generate a desired output; the clarity and specificity of prompts affects the quality of AI responses
As defined in The Economics of Artificial Intelligence | Tyler Cowen
Other definitions
Input text or instructions given to an LLM to elicit a desired response or behavior.
Yann LeCun Says LLMs Have 2 Years Left…
Where it comes up · 4
- This is not the AI we were promised | The Royal Society5 claims
- Yann LeCun Says LLMs Have 2 Years Left…1 claim
- How Investors Are Using AI1 claim
- The Economics of Artificial Intelligence | Tyler Cowen1 claim
Most important claims · 8 of 8Search all claims for “prompt” →
- Large language models work entirely through completion from prompts: given a piece of text, they predict the likeliest next word and repeat this process iteratively to generate responses, exactly like smartphone autocomplete that suggests 'late' or 'in the pub' after 'I'm going to be.'This is not the AI we were promised | The Royal Society
- 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
- LLMs are intrinsically unsafe because they cannot close the gap between training error and test error—there is always a new prompt where they will do stupid things, unlike objective-driven world model systems which can enforce safety constraints by construction through cost functions.Yann LeCun Says LLMs Have 2 Years Left…
- The transformer architecture, published by Google in 2017 in a paper called 'Attention is All You Need,' is fundamentally a machine for learning what word is likely to appear given a particular prompt, and was the breakthrough that enabled practical large language models.This is not the AI we were promised | The Royal Society
- A language model assigns probabilities to sentences by learning the probability that a word will appear given a particular sequence of prior words, approximating these probabilities using enormous neural networks trained on vast amounts of training data.This is not the AI we were promised | The Royal Society
- The speaker and his co-blogger Alex Tabarok have fundamentally changed how they conduct intellectual arguments—instead of yelling at each other, they now send links to different AI models and debate which model is correct, using AI as evidence in philosophical disputesThe Economics of Artificial Intelligence | Tyler Cowen
- The best way to learn prompting is to start with a simple prompt that yields a broad answer, then add layers of detail and observe how responses change, because chat models have few controls and output is sensitive to ordering and context loading.How Investors Are Using AI
- A nutrition chatbot, when asked about food recommendations for an unusual dietary restriction, provided good recommendations, pointing to challenges in deploying AI for sensitive applications like NHS chatbots.This is not the AI we were promised | The Royal Society