> AIs already improve themselves via training Marginally. Model collapse is still a problem. Continuous learning is still a problem. For AI to make a big leap

> AIs already improve themselves via training Marginally. Model collapse is still a problem. Continuous learning is still a problem. For AI to make a big leap we need a big break through.