AI (Artificial Intelligence), most specifically LLMs are technologies used to generate natural language content, allowing users to engage in conversations, ask questions, and receive answers. AI can help you brainstorm and generate new ideas as well as generic text to fill in gaps. It can also be used to edit drafts to ensure that the content is polished and free from errors.
A large language model (LLM) is a language model characterized by its large size. Its size is enabled by AI accelerators, which are able to process vast amounts of text data, mostly scraped from the Internet. LLMs are artificial neural networks which can contain a billion to a trillion weights, and are (pre-)trained using self-supervised learning and semi-supervised learning. Transformer architecture contributed to faster training. Alternative architectures include the mixture of experts (MoE), which has been proposed by Google in 2017.
As language models, they work by taking an input text and repeatedly predicting the next token or word. Up to 2020, fine tuning was the only way a model could be adapted to be able to accomplish specific tasks. Larger sized models, such as GPT-3, however, can be prompt-engineered to achieve similar results. They are thought to acquire embodied knowledge about syntax, semantics and "ontology" inherent in human language corpora, but also inaccuracies and biases present in the corpora.
Notable examples include OpenAI's GPT models (e.g., GPT-3.5 and GPT-4, used in ChatGPT), Google's PaLM (used in Bard), and Meta's LLaMa, as well as BLOOM, Ernie 3.0 Titan, and Anthropic's Claude 2.
Appropedia allows AI-generated content[edit | edit source]
Use of AI on Appropedia is allowed under the following conditions:
- There is not available content on Wikipedia, a different wiki, or any source shared under an open license.
- The content to be created will be used to support your original research.
Dangers of using AI[edit | edit source]
- The information generated by AI is not always be accurate or reliable, and you're still responsible for anything you publish on Appropedia.
- AI's may not always understand the context of a conversation or the nuances of language, so you should review and edit the content generated by AI to ensure that it is accurate, unbiased, and appropriate for the intended audience.
- Remember that AIs are programmed to generate text that makes sense, not necessarily text that is accurate or truthful, so they sometimes hallucinate. Also, they produce content based on the information they have been trained on, so depending on the type of prompt, the information may be outdated and biased.
- Large language models are built using existing materials, including content on Appropedia, sources are usually not attributed or mentioned. This undermines the work of individuals who do this type of work.
- There are inherent biases in AI that can render communities invisible, or perpetuate power imbalances.
Recommendations for using AI content[edit | edit source]
There's no requirement to cite the AIs you use to generate text (but you are encouraged to do so, and may be required to do so if you are editing Appropedia as part of your work, organization, or course), just as there's no requirement to cite the software you used for spellchecking and grammar corrections. However, it is advisable to cite the sources from where the AI generated the text, so users can verify it. Doing so may require research, or you may use AIs like Perplexity that already include the sources used.
We recommend AI for the following actions:
- Create summaries of texts (especially those authored by you or from open sources, including long pages on Appropedia).
- Sumamrize video transcripts to ensure that embedded content such as YouTube videos are accessible offline or more easily understood by others.
- Ask an AI to criticize or raise objections to an essay or piece of content.
External links[edit | edit source]
- ChatGTP - The most well known AI
- Perplexity - AI that generates text while citing its sources