Right. I don’t see why human-level AI is the criteria for AGI. I think we can have weak AGI, potentially long before we have strong AGI. We need clear benchmark

Right. I don’t see why human-level AI is the criteria for AGI. I think we can have weak AGI, potentially long before we have strong AGI. We need clear benchmarks, otherwise it turns into theology.