AI or DIE

[Manlio]

Integrating AI tools into IDEs will probably be the single most critically important innovation for programmers during this decade.

I know I'm an unimportant voice in this community, but for whatever it's worth, I'd like to urge everyone to support the integration of major AI engines into the Lazarus IDE as soon as possible.

A few days ago there was a thread here ("winter is coming" or something like that), and it almost looked like a race of who could find most reasons for shitting on AI and for not wanting to use it.

This reminded me of Robert McHenry, former editor-in-chief of the Encyclopedia Britannica, saying that Wikipedia would never amount to anything because "Wikipedia is like a public restroom where anyone can write anything on the walls."

I sincerely hope that those attitudes are not a representative sample of the FPC / Lazarus community, and especially not a representative sample of its leadership.

Now, of course there's massive hype, of course AI is not going to just replace programmers, and of course to many other reasonable observations - but there's no escaping the fact that this is watershed moment.

A little delay before jumping into action may be OK, if the time is used for studying the landscape and making the best possible decisions in terms of general approach and architecture.

But without decisive action in this regard, the big gap that already exists between the most popular languages/IDEs and Pascal/Lazarus will widen much more and beyond repair. Old programmers like myself may still stick around for another decade or two, and then that will be it.

On the other hand, this could be the moment when the FPC/Lazarus community decides to grab this particular bull by the horns, tame it, and make it work for itself as well as, or better than other IDEs do.

In particular let's remember that a coding AI companion is especially useful and productive for learners and junior programmers. There's a lot of basic and intermediate stuff that young programmers need to learn and master, and this is among the stuff that AI does best -- show, do, and teach how well known things are done. And if I'm not mistaken, young programmers are what the FPC/Pascal community desperately needs if it wants to survive longer than a few more years in the future.

And so forgive me if I sound a tad dramatic, dear brothers and sisters, but I honestly believe that "AI or DIE" is the reality of this historical junction.

Kind regards to all, happy new year to all, unending gratitude to all who helped building these fantastic FPC and Lazarus tools.

Sincerely,
Manlio Mazzon

P.S.

I have been working with Pascal since before PCs were invented. I was there when smart mainframe programmers were skeptical of PCs and didn't quite trust that new Turbo Pascal 1.0 thing to do as good a job as their old trusted IBM Pascal three-pass compilers. Fast forward a little, and I was there again when plenty of "smart" people though the Internet was only a fad. Fast forward a little more, and here we are again, smart people being skeptical about AI integration with programming -- deja vu all over again! No matter how things go, I'll die a lifelong Pascal coder, but what about you and those who will follow? All the best!

[marcov]

Freepascal and Lazarus are volunteer projects. We don't have a lack of policy makers and motivator. We have a lack of people implementing stuff.

And aside from that, IMHO it is too early to invest any scarce resources that FPC/Lazarus has into it. AI on the desktop is all still in the infancy, many of the solutions that are compared with  won't use local AI apis, but instead remote, large server parks full of $2500 accelerators and the business case is mostly ads, which FPC/Lazarus also doesn't do.

But as said, VScode has FPC support and has first class access to copilot APIs and the like. It makes more sense to do the experimenting there.

So a good first step maybe would be to found an ads service to display ads in Lazarus ;-)

[jamie]

Looks like promotional junk mail to me.

Let's not feed the pig.

[Joanna]

Coincidentally I was talking to some people I know in chat recently about the repeated posts demanding that ai be put into the Lazarus ide. One of them offered this advice

Why doesn't he ask his gimpy robot pal to do it if it's so great?

Another friend in chat mentioned that there were a lot of hit and run accounts in forums lately trying to push ai.

I don’t know what all this hype is about but I suspect that someone wants to waste the developers time and get them working on something that has absolutely nothing at all to do with free pascal. I will not use a lazarus ide with ai in it. I will not update if this happens.

In particular let's remember that a coding AI companion is especially useful and productive for learners and junior programmers. There's a lot of basic and intermediate stuff that young programmers need to learn and master, and this is among the stuff that AI does best -- show, do, and teach how well known things are done. And if I'm not mistaken, young programmers are what the FPC/Pascal community desperately needs if it wants to survive longer than a few more years in the future.

What’s with the scare tactics? People should learn and practice programming, not rely on some ai crutch.

Ps. The title of this thread sounds like a death threat !

[marcov]

I don't think the post starter is a bad actor (not with 150 posts to his name, that would be extremely rare). It is just that each and every IT publication is pushing AI as the next best thing to loosen the corporate purse strings. That often becomes an echo chamber.

We have all seen it before with web1/.com,dotnet, singularity , win8 apps, web2, mobile phone apps , cloud and other earlier hypes. Sometimes something happens (like e.g. with the various web frameworks and mobile apps), sometimes not (win8 app platform). But it is quite often different from just copying the approach that big business takes, and more importantly, all predictions that it would overtake the corebusiness haven't panned out.

The first step in any hype is not aimless pushing, but actually trying to imagine how it would fit within the project (or in a 3rd party add-on)

[Joanna]

Ok that explains it. I don’t read IT publications.
I’ve seen otherwise intelligent people get tricked many times in many ways.

I’m getting getting tired of all these redundant threads demanding ai.
Lazarus ide is opensource isn’t it? They should modify their own copy of it if it’s so important to them.

[Curt Carpenter]

Manlio is probably onto something.

I remember when I took a chance at a computer store and bought Delphi 1.0 -- after having put up with writing really lousy programs for windows for a while.  It was magic. 

Maybe an AI assisted IDE would deliver a similar experience,  but I just don't have the imagination to see how that might happen.  I think it will happen though.  I figure I use maybe 10% of the total capability already made available in FPC/Lazarus -- and AI helper might boot me up to 15%.  Probably too late for me, but imagine what programming for pleasure might be in ten or twenty years... 

[simone]

After my degree, about 20 years ago, I got a PhD in artificial intelligence. These academic studies, although a bit dated, give me a perception of A.I. different from that of common people, who probably don't know what they're talking about. I suggest to all those who want to talk about A.I., to read at least one serious book on the subject.

Given the above, I'm still very skeptical about modern A.I., and I don't see valid reasons to invest the few resources of fpc&lazarus in this direction. The priorities should be web and mobile development, which I think are shared by the project leaders.

[marcov]

In addition to what Simone says, I think AI in Lazarus would be more something like SVN or GIT integration in Lazarus, i.e. integration of some external service rather than doing it ourselves.

So more Copilot/Pascal integration than own AI framework. But as said before, that should be pioneered on VSCode to harden the Copilot Pascal support.

[dbannon]

Only five days ago, this story was all over the news services -

The New York Times Sues OpenAI and Microsoft Over A.I. Use of Copyrighted Work

Millions of articles from The New York Times were used to train chatbots that now compete with it, the lawsuit said.

https://www.nytimes.com/2023/12/27/business/media/new-york-times-open-ai-microsoft-lawsuit.html

Sure you want to play in that space ?

Davo

[Joanna]

Ai is probably scraping all the information off this forum and they have enough info to make a counterfeit website or chatbot 

I don’t want any of my conversations or code to be used for financial gain by people creating ai.

After my degree, about 20 years ago, I got a PhD in artificial intelligence. These academic studies, although a bit dated, give me a perception of A.I. different from that of common people, who probably don't know what they're talking about. I suggest to all those who want to talk about A.I., to read at least one serious book on the subject.

I once thought that the name ai meant code that could react to new conditions by analyzing data was so complicated and smart that it was intelligent. Nope nothing like that it’s just a wholesale plagiarism of intellectual property produced by others For profit. They probably don’t even bother to validate if what they are plagiarizing is correct, that would cost too much. Just toss it all in there , gullible lazy people won’t know the difference. Garbage in garbage out!

I remember years back, people were fantasizing about how magical computers would be super smart and know everything . That is not how it works !! Computers are not smart at all. Software is only as good as the abilities and intentions of the people who wrote it.

Please don’t put any of this ai garbage anywhere near Lazarus ide or fpc 

[Curt Carpenter]

OK -- but for what it's worth and from what I've read, the next computer motherboard you buy to run your code will probably be designed with the assistance of some AI/machine learniing-based tools, and generative AI technology is already being used in mechanical design if NASA's Tech Briefs can be believed.   

I think its just a matter of time -- there are some amazingly creative people out there. 

[440bx]

I suggest to all those who want to talk about A.I., to read at least one serious book on the subject.

Is there an entry-level, easy to understand book on A.I that you'd recommend ?

I would love to have the equivalent of "Brinch Hansen On Pascal Compilers" for A.I if such a wonder book exists.

Thanks.

[Leledumbo]

The title is too much, no IDE will die even if they don't integrate any AI into the editor's intelligent system. More mediocre programmers will be coming instead. We've had a generation of programmers who can't code without Google, followed by those who can't code without StackOverflow and this is just the next update to it.

[TRon]

The title is too much, ...

AI or DIE

Artificial Incompetence or Do It Eventually

 

Why bother if MS already mentioned that it wants to integrate it in the OS itself (ASAP) ?

A few days ago there was a thread here ("winter is coming" or something like that), and it almost looked like a race of who could find most reasons for shitting on AI and for not wanting to use it.

The company I work for creates/maintains and trouble-shoot soft and hardware for mission critical production processes, and whether or not you like it or not is bound to government regulations and standards and industry guidelines. AI does not meet these criteria and none of the buzzers keeping the AI buzz alive want to take things like reliability into consideration. You perhaps call that shitting on AI, I would rather name it reality (something that AI has no notion of and never will). Models trained with data generated by humans are biased, racist and unscrupulous.

You can train a model till death and it will still not be allowed to use in our company. It is a legal nightmare no matter if the produced result or assistive technology is actually useful or not.