Gaia whitepaper is out! 🔥📖 Read me: Thread:

Bit of rant here — bare with me plz... 1/ Imagine in today's world, how a small group of people and companies serve AI inference—it's all a walled garden. From training models to load balancing compute, covering infrastructure costs, handling prompts, and even data handling or privacy, it's up to them.

2/ Why is it that I could give all my data, IP, likeness to a GPT, potentially help train the models, develop stronger prompts as a developer, and initiate better applications of the content—yet I don't get to decide how I'm compensated for providing that? I don't get to decide who ALL can run the service for me? Or what kind of service I even need besides which LLM I should choose... there is much more under the hood to decide on...

3/ The AI provider gives me no say on how they manage their compute, or how many tokens I actually need for a response. When the bill comes, I have no input on how much I pay—locked into those 20-30% Web 2.0 platform fees.