Retrieval-augmented generation (RAG) models pull info from specific knowledge bases such as academic and research databases. They have access to current content where as LLS models such as ChatGTP use dated information and are trained with a much broader content range.
| PROS | CONS |
| You know were the info comes from; it is limited. | Cannot breach subscription paywalls. |
| Pulls from open access resources. | Students cannot get a broad view of topics. |
| Fewer "hallucinations." | |
| Help bridge complex ideas. | |
| Content is not dated; it uses current info. |
A Wikipedia-like report on your topic. Different AI specialists investigate your topic; each provides unique insights (e.g. nutritionists, urban geographer, basic fact writer, sociologist, historian). The end product combines all perspectives into one report. Links to information sources are provided. Produced by Stanford.
Perplexity AI serves as a conversational search engine. It responds to user queries using natural language, making it easy for users to obtain straightforward answers. When providing information, it includes inline citations to original sources to back up its responses. It is privately owned.
Semantic Scholar is a free AI-powered research platform that aggregates scientific literature from various fields. It uses AI and engineering to understand the semantics of research papers, helping scholars discover relevant publications in over 211 million research papers. Researchers can use it to find relevant studies efficiently and explore connections in their fields.
ScholarGPT is a specialized version of ChatGPT designed to assist with scholarly research with "built in critical reading skills." It accesses Google Scholar, PubMed, JSTOR, Arxiv, and more. It provide evidence-based information from academic journals, conference papers, theses, and other scholarly literature to over 200 million resources. It does a poor job of providing citations.
SciSpace provides insights and summaries from over 200 million research papers. Users can interact with one specific document at a time to get precise answers. Other resources include an AI writer tool, citation generator, paraphraser, AI detector, and links to other papers in SciSpace.
Connected Papers is a visual tool to help researchers and applied scientists find academic papers relevant to their field of work. analyzes thousands of papers and creates graphs to show their similarity and connections.Large Language Models (LLMs) are good at predicting the next word sequence from a very large knowledge base. They are not as accuracy and reliability as RAGs and have a tendency to "hallucinate" and make things up. LLS models such as ChatGTP are more broadly trained.
| PROS | CONS |
| good for background research. | cannot evaluate credibility of info. |
| natural conversation and easy to understand. | cannot breach paywalls. missing info |
| easy to ask follow up questions. | trained on dated content. |
| opposing view points are readily available. | may "hallucinate" and make things up. |
| help grasp terminology use in fields of study. |