Interesting article about AI

[Bogen85]

With that point we should be still be cave man. Not improving our techniques.

I completely agree.

[gidesa]

In order the AI work fine, I must know how to do the program myself, else how I can instruct it to do it?

But how much time was required to write a very detailed prompt for AI, that generates that good results?
Have you "feeded" AI just with a pre-existing detailed program design document?
Because one thing is to use a document created for other scopes out of AI; another thing is writing that document especially for AI.

[alpine]

Just upload the documentation of the API, the program reads the documents and creates everything for you.

What about the legal issues? You have disclosed the 3-rd party API in a clear text to whom?

This is another kind of payment for the POS app, the app already suppors cash, card, but was needed QR as payment option. The API is third party.

I see, an addition of a coupon payment.
Will it turn out that the work done is minimal and the assessment of acceleration (AI vs. Lainz) is mostly subjective?

[Bogen85]

In order the AI work fine, I must know how to do the program myself, else how I can instruct it to do it?

you already know how to program because you started before ai was invented. Programmers in the future will probably not want to bother learning how to program if they can avoid it. I would never want my code created by a machine that lacks creativity and just regurgitates whatever was fed to it. Then again I’m a hobby programmer and can program any way I like..

Replace AI with FPC and see how that sounds.

You feed code into FPC. It produces boring machine code based on your input. If you feed it faulty code it rejects it if it can or produces faulty machine code if it can't.

Same with AI tools, you are supposed to be the one with the creatively. You take the time to do some new thing without AI, it may take days or weeks that you don't have, and you never get it done. You spend a couple hours getting over the hump with it with AI, and you are days or weeks ahead of where you would have been without leveraging AI. Does not matter if it is for hobby or professional programming.

Just like many programming languages are not for all programmers, neither are AI programming tools.

As far as programmers now or in the future not wanting to learn programming, that kind of thing is and has always been going on.

Each decade from the 40s it has become easier to learn programming. To level the playing field, I'm talking about those with access to computers and being allowed to program if they wanted.

Many who learned programming as a teenager in the 80s would have no desire to learn programming had they been a teenager in the 40s.

If I was 40 years younger and just starting to learn to program now there is a very good chance my programming knowledge and ability would have grown a much faster rate and I'd be much more of an advanced programmer in my 20s then I actually was in my 20s.

And "very good chance" is not a gamble. I've seen this for junior engineers. Online resources, more comprehensive and robust programming language ecosystems, and tools like AI are accelerators in learning, for those that want to learn.

Just like having a complete set of encyclopedias in my house rather than having to go the Library.
Just like having the Internet rather than having to rely on an outdated set of encyclopedias.
Just as those are accelerators for those that want to be accelerated, so are AI programming tools.

To be arguing against AI programming tools is like arguing against encyclopedias, using arguments like, well, you get more exercise walking to the library.
And that is true, you do get more exercise walking to the library.

Just like you get to beat your head against the wall for days trying to get over the hump on some new thing not availing yourself of AI tools rather than hours availing yourself of them.

Are you better off taking days rather than hours? For me, yes. There have been many things in programming I spent days or hours getting over the hump on, only to realize that was not going to be for me in the long run.
Many other times those were for me in the long run, but a lot of time was wasted learning tools, languages, and libraries that were not going to benefit me in the long run.
I'd rather waste a few hours of my time realizing something is not for me, rather than days or weeks.

[lainz]

This is another kind of payment for the POS app, the app already suppors cash, card, but was needed QR as payment option. The API is third party.

I see, an addition of a coupon payment.
Will it turn out that the work done is minimal and the assessment of acceleration (AI vs. Lainz) is mostly subjective?

Is not subjective. I've done same QR code before and it takes time to code and debug it. At lest 16 hours of work or more.

Edit: It includes writing the http connections, parsing the JSON, adding the logic as relationships between these endpoints, coding the ui, translating the ui, testing the ui, testing the endpoints, testing the logic.

With AI I only fixed some small bugs, and I just go to test the app that worked almost at the first time.

[Bogen85]

In order the AI work fine, I must know how to do the program myself, else how I can instruct it to do it?

But how much time was required to write a very detailed prompt for AI, that generates that good results?
Have you "feeded" AI just with a pre-existing detailed program design document?
Because one thing is to use a document created for other scopes out of AI; another thing is writing that document especially for AI.

A few years ago I was at home and not able to go back to work for health reasons for several months (or to even work from home because of the demands and challenges of my job given my health at the time).

Since I had nothing better to do and could not work I took a deep dive for several months into learning AI programming tools.
Just like any other type of programming tool or library it took weeks to get over the hump and months to get really good at it.

To me it was time well spent and when my health situation turned around and I was able to work again I then had a new advanced tool in my programming tool kit that I was skilled at using.

[Bogen85]

This is another kind of payment for the POS app, the app already suppors cash, card, but was needed QR as payment option. The API is third party.

I see, an addition of a coupon payment.
Will it turn out that the work done is minimal and the assessment of acceleration (AI vs. Lainz) is mostly subjective?

Is not subjective. I've done same QR code before and it takes time to code and debug it. At lest 16 hours of work or more.

I agree about it not being subjective.

Subjective maybe in that AI programming tools are not for everything, but definitely not subjective for those skilled in a variety of programming tools including AI based ones.

[alpine]

This is another kind of payment for the POS app, the app already suppors cash, card, but was needed QR as payment option. The API is third party.

I see, an addition of a coupon payment.
Will it turn out that the work done is minimal and the assessment of acceleration (AI vs. Lainz) is mostly subjective?

Is not subjective. I've done same QR code before and it takes time to code and debug it. At lest 16 hours of work or more.

I agree about it not being subjective.

Subjective maybe in that AI programming tools are not for everything, but definitely not subjective for those skilled in a variety of programming tools including AI based ones.

@lainz
Do you count those 16 hours (on AI behalf) as a worktime in which you have learned all the problem details and then asked the AI the right questions? And after that you're prepared to verify its results for correctness?

[lainz]

Good point. Yes I think knowing how an API and QR works is mandatory. So previous knowledge is still needed.

But then it saves you time.

[lainz]

I said it before you need to know how to solve the issue first not leaving that to the Ai alone

[Bogen85]

I said it before you need to know how to solve the issue first not leaving that to the Ai alone

I agree. I've learned when using AI programming tools that if it is not a problem I could have solved myself (days or weeks vs hours or days) then AI is not going to be able to help me.

But I've also learned how to have a short few minute dialog with an AI tool and a few web searches to clarify that for the problem I'm trying to solve if an AI is really going to be able to help me.

If the AI tool can't demonstrate within the first few minutes that it will be able to help me solve the problem (A) I don't leverage it. I either solve the problem (A) a different way or put in on hold and solve other problems (B-Z) until I've learned enough to know how I could solve the problem (A) then I might revisit it using the same AI programming tool or a different one.

It is like any other programming tool or language. Go with what you know, go with one that has the most libraries an ecosystem suited to solve the problem, or learn another language or tool that is more suited. All have trade offs.

[440bx]

To be arguing against AI programming tools is like arguing against encyclopedias,

No, it isn't.  A good encyclopedia is verified for accuracy, AI will blurt _anything_, no matter how preposterous with the "confidence" of a world class expert.

AI is a digital politician, you always get an answer.  Correctness is not a concern.

Owning a calculator (or an encyclopedia) does not make anyone a Mathematician.

[Curt Carpenter]

Meanwhile, from an article in The Conversation (U.S.)

A Swiss church recently installed an AI Jesus in a confessional as part of an art project. Programmed using theological texts, the AI Jesus interacted with visitors through a monitor placed behind a latticework screen.

To clarify, the AI Jesus was not actually hearing confessions. But the act of placing an AI project in the sacred space of a real confessional was “disturbing,” writes Joanne M. Pierce, a historian of Christian worship at Holy Cross. Sins are confessed only within a human community.

[Thaddy]

You're not supposed to use AI if you don't want.

Don't take a shower if it is too hot or too cold...( well, the latter can be skipped).

[gidesa]

To me it was time well spent and when my health situation turned around and I was able to work again I then had a new advanced tool in my programming tool kit that I was skilled at using.

No doubt that knowledge on AI is well worth. And also exciting for people as me, old fan of science fiction and robots :-)
But, in my opinion, it's interesting and important to evaluate the additional effort for using AI, i.e. the effort to writing very detailed prompt.
I have done some experiment, and it's not so trivial.