I like the essay a lot and I find it intellectually amusing that many negative comments here manifest exactly the flaws that could be prevented by learning the essay writing as described in the blog. The article clearly states assumptions, definitions, context and limitations of it's arguments and in that specific area it's very hard to refute them because they are sound. On the other side the gist of many comments here is that article is negative towards GPT and should be positive, i.e. the same generalist aproach that GPT generated text would take and which is hard to argue about.

I'm not going to analyse particular flaws of separate comments because the article itself is really comprehensive and answers for many objections are better in original form there than any reproduction I can make here. But the high number of rebuttals that obviously have problems either understanding the essay or formulating a solid counterargument is curious. An obvious explanation is that some people couldn't resist the pleasure of letting AI dismiss the statement about insufficient AI capabilities, but that's a lazy one.

Comments saying current AI is comparable to actual human inteligence may be right even though I can clearly see that AI is not performing actions I consider necessary for thinking. It's because I consider my own mind as a model of human inteligence, but as I learned many times before, the thinking process may be hugely different for different people. I have no idea if other people also do need to create an explicit model of something in their head to be able to "think" about it.

It would explain A LOT for me if thinking of some other people really works in the same way as in GPT. That situation should however still be viewed not as a technology enhancement so big it reaches human level but reconsideration of what we consider a human level.

This happens to me a lot. I want to make a joke, but it gets so long an complex that it becomes an analysis.