ChatGPT and other recently launched Chat AI tools (such as Bard and Bing) are what are known as “natural language processing” tools (they chat). They use a large language model, or LLM, along with human feedback and training to generate responses. A large language model, or LLM, is a deep learning algorithm that can recognize, summarize, translate, predict and generate text and other content based on knowledge gained from massive datasets.
Information literacy is the skill set required for recognizing the nature of an information need and fullfilling the need by effectively locating and ethically using qualilty, reliable, and applicable information content.
Nothing.
It works entirely on probability of what the next words should be, based on its training data which includes a wide array of human created textual, web-accessible content up to June of 2021. ChatGPT cannot cite the sources from which it extracts information.
Yes, in a manner of speaking.
ChatGPT is only predicting what words are likely come next in response to a user's prompt. It’s been shown to be wrong about 15-20% of the time, sometimes with hilarious or upsetting results. This is known as “AI hallucination.” False information is stated with the same confident tone as factual. ChatGPT often will make up sources if asked to create a biblography.
ChatGPT is good for: Combining ideas, Brainstorming ideas, Modifying written content (spelling, grammar, voice), Breaking writers block, Suggesting thesis statements, Creating outlines, Summarizing text,* Simplifying text*
*Always double-check for meaning.
Images created in DALL-E by Helen Lane
| Have a question or comment about these guides? Contact: libraryreference@fitnyc.edu |
