bbor4 months ago
Thanks for sharing, this is great work! I will say that their analysis is severely handicapped by seemingly being non experts in AI, leading to them missing the big picture issue here that the “Small is Beautiful” cybernetic framework can bring to LLMs: purpose-built democratic FOSS models like Llama vs. massive corporate ones like GPT. It also leads to them making some silly mistakes, like this paragraph;
we only propose to offer a speculative encounter with ChatGPT and DALL-E systems. Although they remain early versions or betas, they have been chosen as they are structured to be infrastructural technologies for cultural production.
That’s pretty understandable tho, and the framework remains useful. I’m looking forward to citing this!