RT by @scite: New on our substack today: we interviewed Josh Nicholson from @scite and uncovered how to combine ChatGPT with scientific literature through scite assistant and redefine scientific citations. Read the rest of our interview here: https://bitsinbio.substack.com/p/interview-with-scite

RT by @montitas: 📣Nova formació sobre intel·ligència artificial: "Como las citas y publicaciones de investigación pueden proporcionar control de calidad a ChatGPT" d' @scite 📅 dimecres 25 d'octubre 🕑15h 💻Virtual (Zoom) ➕ Info i inscripcions: http://tuit.cat/k2dw0

R to @aicrunchio: 2 - Optimize current systems and existing infrastructure. We have a lot of technology which we cannot just get rid of to start from scratch. Thus we rely on #AItools to enhance them and make them work in the most beneficial way for our climate.

R to @aicrunchio: 1 - With AI models for prediction and monitoring, scientists can better understand climate change and what to expect. Meaning, that we can start working on solutions, trying to prevent the disaster. Thus, @scite can provide you with scientific researches and data by request.

R to @domenic_rosati: The last thing I will say is it doesn't really help with pointing to the relevancy issues we are facing - it seems to still really have a bias towards semantic similarity evals which is simple to solve for but doesn't get at our users relvency problems of citation appropriateness

R to @domenic_rosati: I have made a RAGAs fork to remdiate this since their code is tied to sentences but it didn't seem to help... So for now it only really works for Context sentences and doesn't help with grounding in longer form text which is kind of important in scientific RAG systems

R to @domenic_rosati: Context Precision/Relevancy doesn't seem to work very well for longform contexts and outputs. We can boost our scores 0.4 points on an in-house dataset of 200 user questions just by using sentences from papers instead of paragraphs... this seems wrong.

R to @domenic_rosati: Just to follow up on this there are a number of problems I think with RAGAs that make it maybe less interesting... its really easy to just maximize the score on trivial tweeks but im not sure they are meaningful

R to @ResearchHubF: 10. @scite This tool aids in any academic research at a go. It helps in assessing, grading, and evaluating the quality and impact of the research. The tool also produces better visualizations and metrics to discern the citation landscape of a particular paper. Check it out !

R to @Artifexx: Thanks :3 love your posts, keep going

RT by @scite: Descobrir literatura relevante para a pesquisa dá trabalho. Inteligência Artificial ajuda nisso. Abaixo eu vou falar do Scite AI, mais uma ferramenta útil

RT by @scite: Yet general systems like Perplexity and Bing Chat might succeed because they find text from webpages that state the answer. Interestingly the only exception is http://scite.ai assistant , which gets it right too.

RT by @scite: 📢@scite and @Pensoft have partnered to bring contextualised citation data to Nature Conservation articles! 👉Learn more on the @ARPHAplatform blog: https://blog.arphahub.com/2023/09/20/pensoft-and-arpha-partner-with-scite-ai-to-dig-deeper-into-article-citations

RT by @awayusof: Here is my newest workflow for an Effortless Literature Review. It uses @perplexity_ai, @scite, @LitmapsApp, @obsdmd and @teampaperpal. Tutorial with detailed examples & all the tools: (Updated with new AI tools) 👇

RT by @saordon: Here is my method to conduct (and automate) a literature review. Using Google Scholar, @scite , @litmaps , #ChatGPT, @zotero and @obsdmd. Tutorial with examples and best practices: 👇