This article was highlighted for me in a Medium email: congrats! This was a great summary of main thrusts of arguments, and I really appreciate it.

There are a few small things that stood out to me that I disagreed with their depiction, so I figured it might be helpful to start a discussion about them here:

1. To my knowledge, which is for at least the last 6 years where I've been paying pretty close attention to them, OpenAI has always been trying to develop artificial general intelligence. It's not like they "went ahead and made it public that yes, artificial general intelligence is their goal" in response to Elon Musk. It has been their public goal for quite some time.

2. In the repeated prisoner's dilemma ("iterated prisoner's dilemma"), forms of cooperation (e.g. "tit for tat", not "defect") are the best strategy. You reference an article which talks about the prisoner's dilemma, but not about nukes, and you give no explanation as to why the decision-making around holding onto nukes is a one-off repeated prisoner's dilemma situation (in which holding on is the best strategy) rather than a repeated prisoner's dilemma situation (in which actions closer to disarmament are the best strategy), which seems more likely in political scenarios where countries will be interacting over and over again with one another.

(Source: https://blogs.cornell.edu/info2040/2015/09/11/iterated-prisoners-dilemma-and-long-term-strategies/)

Overall, great job! It's nice to only have a few points of obvious disagreement, and for your writing to be clear enough that I can quickly pick up on those points. Thanks for a solid article!