Artificial Intelligence and Lazarus ide

[MassaFernando]

Hello everyone, it's been a while since artificial intelligence has been around, and I've seen little to no discussion directly linking AI with Lazarus. I also imagined that the Lazarus IDE would explode, bringing many new features and surpassing Delphi, thinking that developers would be using AI to evolve the source code. But it seems that everything is the same...
Does anyone know if future versions will feature AI agents assisting in the development of new tools?

[Retrofoxed]

There is now a board dedicated to AI.

As to using AI to improve Lazarus and FPC, there's been a lack of consensus, i.e. some devs are pro-AI and some are against.

As things stand, some use AI and some do not, but I am seeing more commits to main for both FPC and Lazarus - it's being used already to accelerate developmemt.

[Thaddy]

There is already code to integrate AI agents in the Lazarus IDE.
(We have a forum for AI questions like this.)

[PascalDragon]

I also imagined that the Lazarus IDE would explode, bringing many new features and surpassing Delphi, thinking that developers would be using AI to evolve the source code. But it seems that everything is the same...

AI is not the be-all-end-all that many people assume it to be. In my opinion the less AI is used the better especially as it doesn't repel those people that don't want to deal with AI and AI generated code (like me).

[VisualLab]

Maybe the reason for the little interest in using AI for programming in Pascal (or Object Pascal) is that:

• Pascal is easier to learn than C, C++, Rust (Pascal has a clearer syntax) or even Python (Pascal has a clearer syntax and much more possibilities), so few people feel the need to ask questions about simple issues using Pascal (Object Pascal),
• Pascal is not as popular as Python or even C++, so new potential users choose what is promoted by the media, and since these languages ​​create barriers for beginners (C++ and Rust have overly complicated syntax, and Python is primitive), beginner programmers use AI to get a hint or code,
• Pascal (or Object Pascal) programmers know the basics well (language, tools, algorithms, libraries) and can solve simpler or medium-difficulty problems on their own, but IA cannot solve difficult problems yet
• for tasks of medium difficulty you need to buy a subscription, and even then the purchased tokens may quickly run out and you would have to buy a more expensive AI option (and this may not be profitable).

We will have to wait a while for truly useful and relatively cheap AI for programmers.

[440bx]

AI is not the be-all-end-all that many people assume it to be.

It is most definitely not the end all be all but, it can help a great deal.  The most important thing I have found when dealing with A.I is, never ever make the mistake of assuming it did exactly what you told it to do, always meticulously double check everything it did and, when it did exactly what you told it to do, don't presume that's right because we humans share a characteristic  with A.I, we make mistakes too.

As an assistant that saves a lot of "drudge work", I find it very valuable and, just as humans, some  models are definitely more capable than others.

[Retrofoxed]

Maybe the reason for the little interest in using AI for programming in Pascal (or Object Pascal) is that:

Probably that, plus
1) Pascal these days is mainly for hobbyists and many of those are aged 40+ who truly enjoy writing code are less likely to take on innovations.
2) The AI training corpus for Pascal is rather small (the more code AI sees the better it becomes), hence subpar results.

[LeP]

@VisualLab, @Retrofoxed

Sorry for OT, but speaking about Pascal I want to add this:

Pascal is not only about FPC or Lazarus but is about Delphi too.
And in Delphi there are a lot of tools (some integrated in the IDE since some releases) and some from third parties (free and paid).

They covered IDE interactions and external integration with IA too. Some of them may be compatible with FPC/Lazarus.
And lot of "Delpihians" propose some projects made with IA (expecially with Claude) exposing the full process made.
Don't speak about webinar and materials.

I don't think Pascal are for hobbiest: Delphi definitely not, others probably they are.

[mas steindorff]

Probably that, plus
1) Pascal these days is mainly for hobbyists and many of those are aged 40+ who truly enjoy writing code are less likely to take on innovations.
2) The AI training corpus for Pascal is rather small (the more code AI sees the better it becomes), hence subpar results.

1) your showing your green brain cells. you should have used AI before posting that statement.

2) Even using AI for other languages like C and C++, it's answers or often wrong for me. But then again, I don't ask newbie questions others may be using it for.  (Gemini & ChatGPT tested).

[linuxbob]

Maybe the reason for the little interest in using AI for programming in Pascal (or Object Pascal) is that:

Probably that, plus
1) Pascal these days is mainly for hobbyists and many of those are aged 40+ who truly enjoy writing code are less likely to take on innovations.
2) The AI training corpus for Pascal is rather small (the more code AI sees the better it becomes), hence subpar results.

1) Given the power and speed of modern Pascal, and the capability of the LCL, FP/Lazarus/Delphi is far more than for "hobbyists and those are aged 40+". Speaking as a 40+ hobbyist electrical engineer, if I want a hobbyist language, I am choosing QB64 Phoenix Edition.If I am coding serious programs, I am using FPC/Lazarus.

2) AI has access to the full compendium of knowledge of Pascal/Object Pascal/FP. No training needed. Therefore, AI can filter and assemble information adequate to answer specific coding questions. I've tested it, and the results I get are satisfactory.

[tooknox]

Probably that, plus
1) Pascal these days is mainly for hobbyists and many of those are aged 40+ who truly enjoy writing code are less likely to take on innovations.
2) The AI training corpus for Pascal is rather small (the more code AI sees the better it becomes), hence subpar results.

Well 1) is just baseless. You would only find older programmers for languages that are old enough to have them that's a sign of maturity and there are people of all age groups coding in pascal. I am a millennial and just a month back made an internal tool with Lazarus/FPC for a Fortune 500 company.

2) have access to ChatGPT Enterprise and other big boys you can think of (courtesy of the said F 500 company) and never had any issue with pascal.

If you want a RAD tool that's not Microsoft and you don't want to do C++, you have no option my friend Lazarus/Delphi is all there is. And while RAD is great n all, you can do as low level stuff as you want...

[Thaddy]

1) is a misconception for database applications. There is a huge professional market for Delphi for that reason, although it has certainly diminished in size by ~80% since the early 2000ths, because of its former scale - like "everybody" used it - it is still huge.

[gidesa]

I am a millennial and just a month back made an internal tool with Lazarus/FPC for a Fortune 500 company.

This is very interesting to know. Although there is an obvious query: is your the language choice, or a company choice? how much projects that company decide to do with FPC/Delphi?
Because in a big company many developers do they private tools, in whatever language they want. But it's very different, of course, what "official" language a company choose to use in enterprise projects.

If you want a RAD tool that's not Microsoft and you don't want to do C++, you have no option my friend Lazarus/Delphi is all there is. And while RAD is great n all, you can do as low level stuff as you want...

Yes and no. Today many interactive apps are web based. Not only internet/intranet, but also local apps, as VS Code.
So, if RAD is related to Gui, one has many choice (although I think web base local apps are greatly inefficient).

[gidesa]

1) is a misconception for database applications. There is a huge professional market for Delphi for that reason, although it has certainly diminished in size by ~80% since the early 2000ths, because of its former scale - like "everybody" used it - it is still huge.

Maybe "huge" is not the best word to use, compared to other languages.
And surely large part of Delphi market is of limited-size legacy applications. They will die when main developers will retire, and the little companies behind them will close their activity.
Anyway it's difficult to find statistics on Delphi market, to do rational comparison. Same for FPC/Lazarus.

[LeP]

1) is a misconception for database applications. There is a huge professional market for Delphi for that reason, although it has certainly diminished in size by ~80% since the early 2000ths, because of its former scale - like "everybody" used it - it is still huge.

Maybe "huge" is not the best word to use, compared to other languages.
And surely large part of Delphi market is of limited-size legacy applications. They will die when main developers will retire, and the little companies behind them will close their activity.
Anyway it's difficult to find statistics on Delphi market, to do rational comparison. Same for FPC/Lazarus.

I don't know how is the market of Delphi, but for sure is not limited to legacy applications. Of course is "nothing" compared to C# or C++ market, but there are many and many professional apps made with Delphi in every fields.

And like always Pascal is died and will die ... yes of course but Pascal is still there, it doesn't know to be died