Some believe open source AI is a way to break out of the familiar proprietary software quagmire that the technology has predictably fallen into. Hugging Face’s Irene Solaiman and AI2’s Ali Farhadi will discuss this complex issue on a panel at TechCrunch Disrupt 2024 — taking place in San Francisco from October 28-30.
AI may be a very new technology in some ways but in other ways, it’s stuck in the past, specifically in that a handful of decades-old companies are pulling the strings and fronting the cash. But unlike a desktop OS or office suite, the resource requirements of AI models make open source alternatives extremely difficult. What will it take to change that?
To discuss the possibilities and challenges of defining, creating, and providing access to open AI systems, we have leaders of two champions of openness: Hugging Face, which provides open access to models, leaderboards, and datasets, and AI2 (short for Allen Institute for Artificial Intelligence), a research outfit committed to full transparency in its data, training, and models.
Irene Solaiman is Hugging Face’s head of global policy, advocating for and researching safe, open, and responsible AI there and with other tech groups. Ali Farhadi’s AI2 spinoff XNOR was acquired by Apple, after which he returned to lead the organization. Both are proponents of openness and transparency — but both also acknowledge the novel structural barriers facing the embodiment of these principles in AI.
It’s certain to be an extremely interesting conversation between these accomplished AI innovators (and their moderator, yours truly), so be sure to get your Disrupt 2024 ticket and join the AI Stage.
Source : Techcrunch
