It’s hilarious that Musk/Zuck think their platforms are data mines of any value. They’ve poisoned their trillion dollar wells.
Google’s in the process of this too, on a larger scale.
AFAIK most research is leaning into synthetic data generation + runtime RAG anyway. This has tons of problems too, especially when reaching out to the web (for reference) is so screwed up, but it’s a less bad approach.
It’s hilarious that Musk/Zuck think their platforms are data mines of any value. They’ve poisoned their trillion dollar wells.
Google’s in the process of this too, on a larger scale.
AFAIK most research is leaning into synthetic data generation + runtime RAG anyway. This has tons of problems too, especially when reaching out to the web (for reference) is so screwed up, but it’s a less bad approach.