"My current impression of OpenAI’s multiple contradictory perspectives here is that they are genuinely interested in safety - but only insofar as that’s compati

"My current impression of OpenAI’s multiple contradictory perspectives here is that they are genuinely interested in safety - but only insofar as that’s compatible with scaling up AI as fast as possible." I'd rewrite this as "...they are genuinely interested in safety - but only insofar as it doesn't get in the way."