This is a special post for quick takes by Thomas Kwa. Only they can create top-level comments. Comments here also appear on the Quick Takes page and All Posts page.
You are right that I deliberately choose not comment here on the existence of other incidents or what OpenAI should do if they do exist. That would make the statement way spicier. I could disclaim this every time but it would be exhausting and potentially not fit in a tweet.
I do somewhat object to saying I "doctored" anything though, which would imply it is tampered with in order to mislead. Basically every big public statement on AI has had to narrow its scope so that industry people will sign it. Also I mostly actively want people on LW to think about the process by which I decided to post something because this gives them more accurate beliefs.
14Vladimir_Nesov
This kind of thing will have the actual effect of being misleading, at least in the less clear instances, even if there isn't intent to mislead. It shows the lack of an intent to robustly avoid the effect of being misleading. Robustly preventing the borderline statements from being misleading would indeed be either inconvenient or make you stop saying things completely on a wider variety of topics. But that's the tradeoff, narrow intent that doesn't track the consequentialist concerns doesn't save the day.
[...]