I've read a few of Gary Marcus' posts and I like him. My position is much more that the capabilities of current models are good enough for a lot uses. I certain

I've read a few of Gary Marcus' posts and I like him. My position is much more that the capabilities of current models are good enough for a lot uses. I certainly don't think language models are "real intelligence" or that we're anywhere near achieving that, just that excellent search and mashup engines (current language models) are proving to be extremely useful.