Hi Koji, thanks for the video! It's very nice and explicative! I do believe it highlights why having grounding with references is extremely important in the domain of natural language search (chatBot style). I'll share it on my LinkedIn with some considerations, I am inviting others to do the same, thanks for your effort! -------------------------- *Alessandro Benedetti* Director @ Sease Ltd. *Apache Lucene/Solr Committer* *Apache Solr PMC Member*
e-mail: [email protected] *Sease* - Information Retrieval Applied Consulting | Training | Open Source Website: Sease.io <http://sease.io/> LinkedIn <https://linkedin.com/company/sease-ltd> | Twitter <https://twitter.com/seaseltd> | Youtube <https://www.youtube.com/channel/UCDx86ZKLYNpI3gzMercM7BQ> | Github <https://github.com/seaseltd> On Wed, 17 May 2023 at 12:09, Koji Sekiguchi <[email protected]> wrote: > Hello everyone! > > Have you been having trouble expanding your business with dense vector > searches using Lucene/Solr due to the remarkable momentum of ChatGPT > recently? > > In Japan, since the emergence of ChatGPT, many of our clients have > been taking a "wait and see" approach to implementing a dense vector > search engine. > > Of course, I believe that everyone participating in this community is > aware of the respective strengths of ChatGPT and Lucene/Solr. However, > customers who aren't primarily engaged in information retrieval may > surprisingly struggle to articulate these unique features. > > For this reason, my company has created a YouTube video comparing > generative AI chat and dense vector search. I hope it will be helpful > in your business expansion. > > https://youtu.be/q_kPN6zImbY > (Note that in the video, we refer to "dense vector search" as > "semantic search", but don't worry about that.) > > Let's harness the wave of ChatGPT and promote the wonderful > Lucene/Solr together! > > The video is set in a fictional "Large Language Model Library" and > features a narrative of about 4 minutes. The generative AI chat and > dense vector search engine each play the role of a librarian, > responding to library users. The story is divided into three parts: > > Part 1) A librarian, portrayed by the generative AI chat, responds to > a young library user. During the interaction, the AI chat librarian > hallucinates, but the easygoing young man doesn't notice. > > Part 2) The same AI chat librarian responds to a different library > user, a woman. The AI chat librarian provides a flawless response, but > when the woman requests a reference book, the librarian is unable to > meet her request. > > Part 3) Another librarian, played by the dense vector search engine, > responds to the woman. This librarian is able to meet her request and > provides appropriate reference materials. > > We wanted to keep the video as short as possible, so it may seem like > we're only emphasizing the advantages of dense vector search over AI > chat, but of course, we all know that both have their strengths. It's > all about using the right tool in the right place. > > I haven't been able to contribute to the community recently, but I'm > always cheering for the Lucene/Solr community. I'm glad to be able to > contribute in this different way. > > Thank you! > > Koji >
