You have a slightly less great model. Depending on your thesis on how fast AI will advance, that might be minimal, or it might be huge. However, "AI is advancing too fast for people to make obvious efficiency improvements economically worth doing" is rather hard to square with "AI is a lie and will never generate profits."