When many students (and others) think of AI, they think first of generative AI, like ChatGPT, Gemini, or Claude. Keenious uses AI differently than those tools; it does not produce its own words. It uses the words in your query to identify already-published articles that are likely to be relevant to that query. The "query" can be a short sentence, question, or even phrase (e.g., nature vs nurture), or it can be an entire document or PDF, or a highlighted portion of a longer document or text.
Like ChatGPT, Keenious uses a Large Language Model (LLM), but rather than predicting plausible text, it uses the LLM to create a kind of "fingerprint" based on themes and topics it infers based on your query, and matches that fingerprint to similar "fingerprints" it has created from the articles in the database of articles and article abstracts it uses, OpenAlex. If your query text is not in English, it will create an English translation (internally) and work from that. Keenious is good at finding relevant articles, even if they don't use the same words you did to describe concepts. It is more subtle than a traditional keyword search.
