You are partially right — OpenAI is way ahead of everybody else. Even though OpenAI team is thinking and doing everything for safe deployment of (baby) AGI, pub

You are partially right — OpenAI is way ahead of everybody else. Even though OpenAI team is thinking and doing everything for safe deployment of (baby) AGI, public and experts don’t think this should be effort lead by single company. So Google naturaly wants to be the counterweight. (Ironic that OpenAI was supposed to be counterweight, not vice versa.) However, when you want to catch up somebody, you cheat. And cheating with AI safety is inherenťy dangerous. Moratorium for research and deployment just doesn’t make sense from any standpoint IMO. Regarding the hand-holding: As Hinton noted, simple extrapolation of current progress yields models that are super-human in any domain. Even if these models would not be able to access Internet, in wrong hands it could create disaster. Or even in good hands that just don’t anticipate some bad outcome. Tool that is too powerful and nobody tried it before.