- Large Language Model (LLM)
- A type of AI trained on vast amounts of text to predict and generate human-like language β the technology underlying ChatGPT.
- Prompt
- The text input a user types into ChatGPT to instruct it what to produce β the quality of the prompt directly shapes the quality of the output.
- Hallucination
- When an AI model generates a confident-sounding response that is factually incorrect or entirely fabricated.
- Token
- The unit ChatGPT uses to measure text length β roughly 0.75 words per token. Context window limits are measured in tokens, not words or characters.
- Context Window
- The maximum amount of text ChatGPT can hold in a single conversation β once exceeded, earlier parts of the conversation are no longer considered.
- System Prompt
- An instruction given at the start of a conversation that sets ChatGPT's persona, tone, or constraints for all subsequent responses.
- Temperature
- A setting that controls how predictable or creative ChatGPT's responses are β lower values produce more consistent output; higher values produce more varied output.
- GPT-4 / GPT-4o
- The current flagship model versions from OpenAI, offering stronger reasoning, longer context windows, and multimodal capabilities compared to earlier versions.
- Multimodal
- The ability to process and generate multiple types of input and output β including text, images, and audio β within a single AI model.
- Zero-shot prompting
- Asking ChatGPT to perform a task without providing any examples, relying entirely on its training to interpret the instruction correctly.
- Few-shot prompting
- Providing one or more examples within the prompt so ChatGPT learns the desired format or style before generating its response.