Leximancer

We’ve been very busy here at Leximancer, not only are we seconds away from #betatesting our #betterthanrag product #Cassandra we have begun selecting features for inclusion in Leximancer 6, our most powerful version yet! We will set up a process to hear from you what your favourite features are and what you hope to see.. #feedback #ai #rag #llm #thematicanalysis #developmenttools #qualitativeresearch #womeninstem

What developers are saying… “How Cassandra is different from AI21 Labs (and RAG in general) is it removes 3 things: 1. the document splitting (chunking), 2. embedding and 3. vector DB steps. Cassandra flow: User uploads a document Cassandra API creates knowledge graph of document content ---- now the knowledge has been stored ---- User asks a question. Cassandra API retrieves top source document chunks that are relevant The LLM now has the users question and the relevant text chunks from Cassandra in its context window LLM generates an answer We can demonstrate that the retrieval of info with Cassandra is as good as traditional RAG (hopefully better) this means that we can do 3 things to traditional RAG processes: - reduce time to answer/upload, - reduce costs (mainly embedding API costs and Vector DB running costs) and - reduce compute required.” #ai #rag #LLM #newmodel #machinlearning

Link ready, https://lnkd.in/gA2rjK5T

Anyone keen for some AI beta testing? Register here https://lnkd.in/gA2rjK5T #AI #machinelearning #llm #Betatesting

Is RAG the answer? Or can we do better?

Confabulation is definitely my new word of the week... #hallucinations #getsomeguardrailsalready

Do you know what a halluncination is? If you're using AI, it's wrong 27% of the time. How do like those odds? #trumplawyer #whoops #hallucinations #ai #chatgpt #guardrails #LLM #security #IPrisk #devtools "In the field of artificial intelligence (AI), a hallucination or artificial hallucination (also called confabulation[1] or delusion[2]) is a response generated by an AI which contains false or misleading information presented as fact.[3][4][5] For example, a hallucinating chatbot might, when asked to generate a financial report for a company, falsely state that the company's revenue was $13.6 billion (or some other number apparently "plucked from thin air").[6] Such phenomena are termed "hallucinations", in loose analogy with the phenomenon of hallucination in human psychology. However, one key difference is that human hallucination is usually associated with false percepts, but an AI hallucination is associated with the category of unjustified responses or beliefs.[5] Some researchers believe the specific term "AI hallucination" unreasonably anthropomorphizes computers.[1] AI hallucination gained prominence during the AI boom, alongside the rollout of widely used chatbots based on large language models (LLMs), such as ChatGPT.[7] Users complained that such chatbots often seemed to pointlessly embed plausible-sounding random falsehoods within their generated content.[8] By 2023, analysts considered frequent hallucination to be a major problem in LLM technology, with some estimating chatbots hallucinate as much as 27% of the time[9][10] and a study finding factual errors in 46% of generated responses." https://lnkd.in/dJJ7VytN)

We are lucky enough to call Professor Hugh Possingham a friend, he is also a incredibly kind and generous collaborator who always has time for random tree, bird and life questions. Thank you for all that you do for all of humanity and the research community Hugh, it is a privilege to know you. Keep up the incredible ethos 😎 💪 it's contagious https://lnkd.in/gZaQjNFN #extinction #qualitativeresearch #ecology #environment #maths #womeninscience #mentorship #ai #socialresearch #LLM #biodiversity #pleaseactnow #epbc

#qualitativeresearch but make it smart not hard. Come and join the team and we can show you how. Survey analysis jump started.

Really valuable in this explosion phase to hear from users of #AI, our goal is to design products that enable high quality research that is grounded in solid research methodology, can be validated and stands up to the rigours of the academic process, an ever greater challenge in the world of #largelanguagemodels and #trainedai that doesn’t represent all of humanity. #unsupervisedlearning #humanbias #qualitativeresearch #qualitativeanalysis #sentimentanalysis #literaturereview #surveyanalysis #socialmediaresearch