LLM Avalanche: Nick Vidal: Open Source AI
Recording: LLM Avalanche: Nick Vidal: Open Source AI
thank you how many people here like sloppy Source raise your hand go whoop his horse yes so my name is I'm part of the OSI the open source initiative now we coined the term open source 25 years ago this is a huge celebration across the world because open source powers everything today and there has been a lot of discussion about open source AI what's actually open source AI if you search on Google open source AI you're going to give 400 000 results there is the AIX that's been drafted in Europe laws that are governing Ai and they have a special Clause around mobile Source AI but we don't know exactly what's open source CI so those sign together with a whole bunch of organizations is startups are trying to work together to create an open process to Define what's open source Di it's an ambitious goal we're going to be participating in several events so here in San Francisco just a week ago we had a meeting at the Mozilla headquarters where a whole bunch of organizations came together to start creating the first drafts of what's open source AI we're going to be participating several events during the year so one of those events is fosse which is an event in Portland we're also going to be participating at a lineage foundation events across Europe in Geneva in Bilbao Spain and we're also going to have online webinars and right now we have a call for papers anybody who wants to talk about open source AI if you have an idea if we should discuss or what's open source to you the call for paper is open and here is the cover proposals so here are some suggestions that we have so one Association what are the basic principles that Define open data Ai and ml another one what's copyrightable are the inputs how brightable are the outputs copyrightable how can we give control over the data the weights the models to businesses and to individuals and how can this help them what's openness and what's transparency of the trading data how about data privacy when we use openai we're basically reaching out on an API and giving our data to open AI how's that if you're a business and you have very sensitive data you're basically giving the everything away and there's security concerns around them also a question about Fair competition right now uh a lot of companies are trying to regulates AI if people want to experiments and try different models and study and build on top of of those models open source and you have to create a Fair competition as well how can we promote commercial growth of Open Source AI also we can discuss the climate impacts of training those large models instead of using large models if we can work those small models and iterates over and over maybe we can get better results than a larger model also responsible AI this has been a big discussion as well how can we promote responsible AI how can we create safeguards for open source AI now finally how can we protect Society from AI so those are some ideas for the common proposals and I invite each one of you if you want to participate in those online webinars or it should be part of the conferences you're more than welcome to to join so with that thank you very much and go office hours oh yeah