Open and Trusted AI: What Do We Need to Make it Real?
Pre-registration by Feb 4 required!
Please note: the venue requires a list of all registrants no later than 24 hours prior to the event. Unfortunately, they will not be able to register additional attendees after this time. Please make sure to RSVP by Tuesday, February 6 at 7pm.
Agenda
19:30 – Arrival
20:00 – Welcome & announcements
20:05 – A brief introduction to the AI Alliance
20:10 – Lightning talks
Observatory of Trustworthy AI Adoption in France – Roxana Rugina (Executive Director, Impact AI)
The Future Needs Open Innovation in AI – Sean Hughes (ServiceNow)
Building the Future of Open, Trusted Data for AI – Sean Hughes (ServiceNow)
Fast-LLM – Scaling Large Language Model Training with Speed, Efficiency, and Openness – Sean Hughes (ServiceNow)
Topic TBD – Joe Olson (IBM)
20:45 – Socializing
21:30 – Closing
Co-organizers
IBM
Meta
Impact.ai
DataCraft
The Future Needs Open Innovation in AI
The rapid evolution of generative AI (GenAI) raises fundamental questions—not just about what AI can do, but about who is building it and for whom. Today, AI development is increasingly concentrated within a handful of companies, creating closed, proprietary systems that limit transparency, oversight, and fair competition. First we will explore why open innovation—collaborative, transparent, and community-driven AI development—is essential for progress. We will discuss initiatives like BigCode, which foster open-access dataset and model development, independent audits, and responsible AI governance. If we want AI that is safer, more innovative, and accessible, open collaboration is the way forward.
Building the Future of Open, Trusted Data for AI
AI is only as good as the data it’s trained on. However, provenance, governance, and licensing issues continue to create uncertainty in AI development. The Open Trusted Data Initiative (OTDI), led by The AI Alliance, is addressing this challenge by building a catalog of fully traceable, transparently governed, and permissively licensed datasets for AI model training. This talk highlights the importance of trusted data in reducing bias, improving model quality, and ensuring compliance with ethical and regulatory standards. We will explore recent contributions from industry leaders like Meta, ServiceNow, AItomatic, PleIAs, BrightQuery and Common Crawl, and discuss how researchers, developers, and organizations can participate in this open-data movement to fuel safer and more reliable AI systems.
Fast-LLM – Scaling Large Language Model Training with Speed, Efficiency, and Openness
As AI models continue to grow in size and complexity, training them efficiently has become a major challenge. Fast-LLM, an open-source library developed by ServiceNow Research, is designed to make large language model (LLM) training faster, more scalable, and cost-effective. This session will provide an introductory overview of what Fast-LLM is, why it matters, and how it helps AI researchers, engineers, and enterprises optimize their training processes. We’ll explore its key advantages—speed, flexibility, and seamless integration—and highlight how it’s already being used in real-world AI projects like StarCoder2. Whether you’re new to large-scale AI training or simply curious about advancements in AI infrastructure, this talk will give you a high-level introduction to Fast-LLM and its potential.
About the AI Alliance
The AI Alliance is an international community of researchers, developers and organizational leaders committed to support and enhance open innovation across the AI technology landscape to accelerate progress, improve safety, security and trust in AI, and maximize benefits to people and society everywhere. Members of the AI Alliance believe that open innovation is essential to develop and achieve safe and responsible AI that benefit society rather than benefit a select few big players.