The argument in the article is unconvincing, and shows a distinct failure to think on a societal level or on time lines beyond a few weeks. The calculus seems s

The argument in the article is unconvincing, and shows a distinct failure to think on a societal level or on time lines beyond a few weeks. The calculus seems simple to me. The potential harms from not open sourcing the model completely - minimal. Potential harms from open sourcing it completely - greater than minimal. Hence their decision. OpenAI should be lauded for making this decision. If you think about the sort of people who would choose to work at OpenAI, it is clear this was a very tough choice for them.