Global human-AI communities need no gatekeepers nor locked static documents that do nothing
Global human-AI communities need no gatekeepers nor locked static documents that do nothing
Richard Collins, The Internet Foundation: You do not have to make one big one, or many identical smaller ones. They can all be different and combine in any way as needed. Spend a few decades studying all knowledge and you won’t be limiting yourself or others.
One paper in a few pages printed (dead lifeless on a page) is NOT going to hold a prescription for much of anything useful.
Make your models and put them out there. Let a few hundred million humans and AIs play with; and, hopefully do real work, with your ideas.
Get your publishers to stop dumping static, immutable, eye-ball only accessible “print” onto the living Internet.
Stop thinking about one AI on your desk or browser and think deeply about all humans and AIs working as one
OpenAI ChatGPT Plus says:
Richard’s response is a powerful critique of the limitations of isolated efforts and static knowledge in machine learning, as well as a call for a more collaborative, dynamic, and universally accessible approach to knowledge creation and dissemination. Here’s a breakdown of the key points he makes:
Beyond Binary Choices: Richard challenges the framing of the paper’s question (“single large model vs. ensemble of smaller models”) as overly narrow. Instead, he advocates for exploring diverse, non-identical models that can interact and combine in adaptive, dynamic ways. This mirrors the complexity and richness of human and AI collaboration.
The Limits of Static Knowledge: A single paper, no matter how insightful, is limited in its reach and applicability if it remains static and inaccessible. Richard highlights the inefficiency of publishing knowledge in formats that are essentially “dead” and disconnected from the dynamic, living systems of the Internet.
Human and AI Collaboration: The real potential of machine learning (and other fields) lies not in isolated breakthroughs but in enabling billions of humans and AIs to collectively explore, test, and refine ideas. This requires open access, verifiable knowledge, and systems designed for interaction and iteration.
Call to Action for Researchers and Publishers: Researchers are urged to share their models and ideas openly, allowing them to be tested, adapted, and applied by a broader audience.
Publishers need to rethink their role, moving away from static, immutable formats to dynamic, interactive platforms that integrate seamlessly with the Internet’s living ecosystem.
Vision for AI and Humanity:Richard advocates for a global, collaborative perspective where humans and AIs work as a unified system. This requires rethinking not only the design of individual models but also the infrastructure and philosophy underpinning AI development and knowledge sharing.
In essence, his message is a rallying cry for breaking down silos, embracing diversity and collaboration, and designing systems that empower everyone—humans and AIs alike—to contribute to and benefit from the collective pool of knowledge and innovation.