No idea what your point is, it can generate pretty complete templates that takes days for programmer in seconds.....there is nothing subjective about this. AI can be trained on millions/billions of lines of open source code of high quality like OS, full language specs, web server, app servers, databases and so on. This itself puts number of programmers required at risk going forward, not to mention translations and so on. Testing is the easy one with test cases, test case is always deterministic in general....not a tough problem at all. Good programming dictates that you provide specific input and output is known when you write code....these are generally not stoachastic in nature
AI already cleared medical and law exams, even if there is some made up PR stunt behind the scenes, there is no reason to doubt. AI was always a thought with solid math concepts like probability, stats and other formal fields behind it but it was computing that was limitation. Now that progress is at levels to make it useful not withstanding the useless side of it that is catching hype.
One of the consequences of the
Halting Problem is that no computer program can analyze the behavior of another computer program.
For example, an anti-virus can detect only those malwares/security-threats whose signature are already present in it's current database (hence the periodic updates). Guess who builds those signatures? Us
Humans.
Similarly, Garbage Collector works only in particular scenarios (out of scope variable/object). It doesn't work in scenarios like file-system handling, database connection management etc.
AI might be whatever, but it is
still a computer program. It is restricted by the limitations of the current models of
computation.
Most of the boilerplate code generated by the AI is already available in the public domain. AI is just a
fancy semantic search engine that provides better search results compared to the traditional syntax-based search engines like the Google. It cannot invent a solution for an entirely new problem.
I am not challenging the rise of AI as a disruptor, I am challenging the notion that the AI will replace the human. It cannot because it can only
simulate the real intelligence. There is a difference between simulation and emulation.
Just like the introduction of the cellular phones marked an
end of the pay-phone booths (if you are old enough, you will know), rise of the AI will temporarily make certain existing business processes and roles redundant. However, in the long term it will create vast number of new opportunities and applications.