LIN only - terrytaylorbonn/auxdrone GitHub Wiki

25.0126

Content for LinkedIn article only

The latest generation of "AI" (search and code-complete) tools will change SW dev. The number of dev workers will increase (not decrease) as SW dev tools go through another evolutionary iteration (a process that has been going on for decades; the rapid decline in assembly level programmer jobs decades ago did not lead to the end of SW dev jobs). Human activity in SW dev will move to a higher abstraction level with higher levels of productivity and employment .

2 AI is starting to become "copilots" and enablers

I’ve written a lot of Api docs, I wrote docs that really show how to use (NO TDLR). I could do that because

  • critical company experts help (minimal required, but critical at times),
  • google, StackOverflow, YouTube videos
  • search algorithms to find by keyword

But the above do not

  • answer questions, or
  • tell how to modify my existing code (the details).

Copilot does. I am not a full stack dev. But I was able, without human assistance, to get this running because of AI. Most interesting, though, was just how much I enjoyed "chatting" with Copilot while trying to implement things. I gradually started to realize I was talking with a librarian, a very capable one, who was steering me to the sounces of info I reuired.

image image

An example is an aspect of copilot demo’s often in doc #238 where I ask “COP[4]: is there yet another way to post?” (“COP” = copilot). Copilot analyzes my code and gives me the code to change/insert. These bots are getting better and better at recognizing the human world.

3 Recent PRIVATE sector investment confirms this

The recent $500 billion private sector AI investment announced by Trump will be money well spent (better than most gov-run uni courses). On a giant card catalog with librarian bots. Tools that will drastically raise productivity. Enable humans to do more than they ever thought possible. I never imagined that that these tools would enable me to get up to speed so quickly in subjects that I was not an expert in. Without human assistance.

image

4 Getting smart about AI

AI is actually mroe interesting once you understand it enough and can get past the hype (and scams). A few of my own convictions about this tech that many think will one day replacement all human workers (something the likes of Gates and Zberg say to get attention).

meanwhile, in gov-run tech education ///// A lot of professionals need AI // is better than gov-run education

image

I speak sevearl languages. And AI tools (google xlate, text to speech, auto xlate in youtube, etc) have made the university style classes obsolete. You can learn more at home (if you have the initiative).

This is also true now of tech courses. after a few decades I took a sw dev course in fall 24 at a local college. It was a major disappointment. Focus on low-level coding, disorganized, no big picture, but about all they frowned heavily on using AI. They wanted you to stay in the same level of code dev (makes sense, easier for a gov-funded organization to teach).

AI has no intelligence

image

I also gradually realzied the librarian i was talking to was not intelligent. Human mind = intelligent.

Computers are dumb binary machines and so is any app that runs on them. Dev tools: first compiler = big deal. Search engines, code complete. And now AI (SI)

artificial Sweetener created a sweet taste. Artificial colors, flavors actually created colors and flavors. But AI only simulates intelligence (SI).

AI is a marketing gimmick that has been used for decades. I remember 40 years ago supposedly brilliant programmers seriously claiming that their primitive binary computer programs were intelligent and could one day replace humans. 40 years ago the movie Tron was based on some bizarre idea that computers back then were intelligent enough to “digitalize” a human that would then live in the computer system. It was good marketing hype. This is what the current AI hype is . hype that sells, makes profit.

AI “training” = building a card catalog

image

Training = misused ai term. Ai does not learn. Its catalogs, that’s why it needs huge data centers. Huge amounts of “training” (I’ve studied it a bit).

Its basically an extremely large and complex card catalog. Soemthing a human could never manage. https://en.wikipedia.org/wiki/Library_catalog