Welcome back! Google just dropped a compression algorithm that could fundamentally change the economics of running large language models. The AI industry's favourite AGI benchmark got a brutal reset, Cursor got caught being less than transparent about where its new model came from and Reddit is trying to figure out how to keep humans in charge of its platform.
In today’s Generative AI Newsletter:
Google: Can TurboQuant really cut LLM costs overnight?
ARC Prize: What happens when every frontier model scores below 1%?
Cursor: Why is nobody talking about the Kimi connection?
Reddit: Is the dead internet theory already here?
