I feel like the models are already smart enough to be called something like an early version of AGI. It's the tooling around it that needs to improve. For exam

I feel like the models are already smart enough to be called something like an early version of AGI. It's the tooling around it that needs to improve. For example, Codex/Claude should be able to code and "see" its changes. If it's developing a mobile app, it should be able to view the real app it's building and then make changes. If it's a website, it should see the changes being rendered and then make further changes if needed. The tooling is well behind model intelligence.