Are we in the first stages of RSI?

In my opinion we are. Yes humans are still in the loop, but models are good enough now to create synthetic data to train better models, and many aspects of curation can be automated now too. Yes it’s not fully automated, but I think the general principle of a previous model making a new model better is holding true. I also found it interesting that Astra recently was able to successfully edit the raw weights of Qwen 3.8. Though the exact criteria of RSI seems to be murky similar to AGI. So I’m curious what other people’s thoughts are.