Tom Barnes comments on What is the current most representative EA AI x-risk argument?