>But nonetheless, they think they have a clear shot to make programming a non labor intensive activity like farming is today.

Here lies the difference between agriculture and software.

Agriculture is there for the sole reason of "feeding people". The demand is capped. Once you have successfully "fed everyone", all the rest of agriculture is luxury.

On the other hand, what I see on the "demand side" of software, is a truly bottomless hole. After the current version of the software, there is a need for the next version of the software, with new and more and better features... Either you do it first. Or your competitor will.

It doesn't ever end.

When you can write double the code in half the time? I suspect that the result will not be: "We will write the same number of lines, and hire fewer people", but: "There shall be four times the amount of code written, beating everyone else to release!"

After all, even currently you always have the choice: Do you finish your software with half the people, in double the time? Or does it pay to hire someone, in order to get out more features, faster, better?

As I understand it, most software development is not "a single person, working long", but "many people working fast". So I see AI as shifting the balance toward "equally as many people, working all that much faster"

>It is at its heart a bunch of nodes linked together with different weights that basically just figures out that “these words tend to follow these other words”

You have just described my brain :D

>It doesn’t use logic in the human sense, and doesn’t have common sense.

I would argue: Neither do we.

>One of the problems is that it actually is has a tendency to make false statements and logical fallacies in its results (while sounding super confident)

I know that problem from somewhere. It did not originate with the invention of chatGPT.

Please be aware that I am not entirely serious... But all of those are also pretty good arguments to deny humans the term General Intelligence :D