Winter is coming (AI-assisted coding and its influence on Free Pascal)

[domasz]

However I make good use on AI tools to overcome one of the biggest issues I have with Lazarus and Pascal - the imo really lousy documentation.

This is surprising for me. I always though the documentation (especially in Delphi) is great and there are a lot of examples.
Install a Delphi if you don't have one. Can even be something old and try its help files. Maybe you will like it more.
When googling code examples use delphi, not fpc, pascal or lazarus, eg. "delphi windows screenshot".
The problems I have is with documentation and examples from third party libraries but then there is this forum.

[cpicanco]

phind suggested to use TMemoryStream instead of BlockRead and BlockWrite directly. Much better.

Code: Pascal  [Select][+][-]

1. function TConsecutivesCounter.ToBinary: TBytes;
2. var
3.  Stream: TMemoryStream;
4. begin
5.  Stream := TMemoryStream.Create;
6.  try
7.   Stream.WriteBuffer(FCount, SizeOf(FCount));
8.   Stream.WriteBuffer(FConsecutives, SizeOf(FConsecutives));
9.   Stream.WriteBuffer(FMaxConsecutives, SizeOf(FMaxConsecutives));
10.   Result := Stream.ToBytes;
11.  finally
12.   Stream.Free;
13.  end;
14. end;
15.  
16. procedure TConsecutivesCounter.FromBinary(const Data: TBytes);
17. var
18.  Stream: TMemoryStream;
19. begin
20.  Stream := TMemoryStream.Create;
21.  try
22.   Stream.WriteBuffer(Data[0], Length(Data));
23.   Stream.Position := 0;
24.   Stream.ReadBuffer(FCount, SizeOf(FCount));
25.   Stream.ReadBuffer(FConsecutives, SizeOf(FConsecutives));
26.   Stream.ReadBuffer(FMaxConsecutives, SizeOf(FMaxConsecutives));
27.  finally
28.   Stream.Free;
29.  end;
30. end;
31.  

[Joanna from IRC]

I tried some AI tools, including ChatGPT, and the code was shit. At first it looked like it saved me some time but results were wrong, the code was slow or broken in just some cases.
I am really surprised any of you find any value in those AI tools.

I agree.  The last thing fpc needs are mysterious libraries doing who knows what.. I have absolutely no sympathy for people who refuse to learn pascal or help others who are trying to learn pascal.
As for documentation , more live humans who like to talk about pascal are needed. I have no interest whatsoever in replacing what few humans that remain with Ai  machines. What a horrible idea. 

[cpicanco]

I tried some AI tools, including ChatGPT, and the code was shit. At first it looked like it saved me some time but results were wrong, the code was slow or broken in just some cases.
I am really surprised any of you find any value in those AI tools.

I agree.  The last thing fpc needs are mysterious libraries doing who knows what.. I have absolutely no sympathy for people who refuse to learn pascal or help others who are trying to learn pascal.
As for documentation , more live humans who like to talk about pascal are needed. I have no interest whatsoever in replacing what few humans that remain with Ai  machines. What a horrible idea. 

I agree 100%. I refuse to replace humans on common tasks. Also, I refuse to neglect how these same human beings can be empowered by AI-assisted tools. Again, tools:

1) Tools that can help speech impaired people to talk again.
2) Tools that can help disabled people to walk again.
3) Tools that can help people working alone to get stuff done faster.
4) Tools that can help socially proficient people to comunicate better and get stuff done even faster.
5) And so on.

[Joanna from IRC]

This ai stuff just seems like a bunch of hype and I certainly don’t want it incorporated into my fpc compiler.

Am I supposed to believe that people are creating all this freely accessible ai stuff out of the kindness of theirs hearts with no ulterior motives? They would never try to manipulate public opinion or insert advertising and spyware would they ?? Or how about claiming ownership of everything you make with their ai ?

If you want to create free pascal apps to do amazing altruistic things more power to you. Surely you will find a way with or without ai .

The fact that there are so few people who are willing/able to have real-time conversations about pascal really distressed me. Have people forgotten pascal so much that they must rely on reference books to remember how to write code? Less than ten years ago it was easy to find people to chatting about pascal. Now it is impossible. Everyone I know in real life scorns the pascal language. Same with people I encounter online.

There is no human endeavor of any consequence that was ever successful without some form of real-time communication. Pascal development is no exception. Without practice, the proficiency for pascal with atrophy just like any other skill. Being unable to think of code well enough to carry on a conversation about it is the first step towards losing the skill to code at all.

I think it’s time we put a stop to this trend of being dumbed down with deceptively “helpful” technology. If anyone wants to discuss this with  me further Please do.

[MaxLemberger]

I can sign that. Integrated in the compiler - that would be a worst case for me. If you see it as a support in the same way as a discovery while searching on the internet, it's ok with me. There's nothing like communication from person to person - that's certainly the best way to learn.

The background is also that people have forgotten how to undergo the effort of the level so that they can get to the place of ascension. The AI should take care of that? Never. In order to be able to program well, you simply have to learn and try things out.

There are simply too many people who think they have gears in their brain and if they use them too much they wear out prematurely ;-).

lazarus.intern.es
lazarus.intern.ws

[MarkMLl]

Unusually, I'm in full agreement with Joanna.

I must admit that I'm quietly amused by this thread's subject line: I'm old enough to remember the term "AI Winter" meaning something very different.

MarkMLl

[lainz]

Some business people are turning the use of AI-assisted tools mandatory. Some are even threatining to fire people that don't known how to use it.

My boss will fire the whole team if there is even a suspicion of someone in the team having used any AI assistant. It is a legal nightmare.

2 cents and just saying.

In my job I must use chatgpt, to solve problems that requires a lot of google search (even with that was not possible with just Google due to lack of documentation) for Android Development, imagine for other less documented areas of programming how useful it is.

Edit: try it yourself, try to make an android application with room database, that has a reactive recyclerview.

All youtube videos use internet in their demos, not room database.

With chatgpt I asking one and other and again and again, I was able to produce all the code without writing a single line.

Also is usefull at debugging, you paste the error and stack trace there and it gives you hints on how to solve it, if not directly the solution.

[cpicanco]

Unusually, I'm in full agreement with Joanna.

I must admit that I'm quietly amused by this thread's subject line: I'm old enough to remember the term "AI Winter" meaning something very different.

MarkMLl

AI summer has come. It's here. How long will it last this time?

[Joanna from IRC]

The background is also that people have forgotten how to undergo the effort of the level so that they can get to the place of ascension. The AI should take care of that? Never. In order to be able to program well, you simply have to learn and try things out.

Unfortunately most people want to be “spoon fed “ or copy from others.

I can also foresee the makers of ai deciding what things they want to help with/promote, thus gaining control of all technology. Everything that can’t be done with ai will simply be disregarded. It’s the consolidation of knowledge. Don’t bother to learn how to do anything Just ask a computer.

Even if ai is a useful technology that could help everyone, it will be hijacked and turned into a tool of of centralized control just as Television and radio were. The majority of Media is owned and controlled by very few people. The results of billions of people being brainwashed by their TVs is not good.

We really need to get back to having real-time conversations with live people about things we care about while it’s still possible and fight hard against anyone who tries to prevent us from gathering and talking.

[440bx]

We really need to get back to having real-time conversations with live people about things we care about while it’s still possible and fight hard against anyone who tries to prevent us from gathering and talking.

I really don't think much of A.I, but I do believe that for a very small number of things it might make a positive contribution, for instance, what @lainz mentioned in his post.

I also think that real time conversations are not very likely to solve the problems we face today.  Real time conversations are happening non-stop everyday.  The so called "smartphones" (I guess the "smart" part comes from the very likely possibility that in many cases the phone is smarter than its user) have made such conversations not only possible but a real problem for the proper functioning of many societies (texting and driving don't go together, nor do texting and _mindlessly_ walking on a sidewalk.)

Personally, I believe the root of the problem is the severe lack of analytical skills in a very significant percentage of the population. Combine that with the desire (and sometimes demand) for instant gratification and there is the making of a severe and chronic problem.  Unfortunately, A.I is more likely to worsen this situation than improve it.  From what I've seen so far, A.I promotes intellectual inactivity (A.I will figure it for you...)

The conclusion is a paraphrase of one of Einstein's well known saying... we need to teach people how to think and, at least occasionally, engage in the activity.  Not only would that be good for programming, it would likely be quite good for any society's health.

[jamie]

I'll side with Joanna, I for one don't want to have to answer to the likes of a Terminator cyborg and have it tell me that "I'll be back"

Jamie

[Joanna from IRC]

@440bx you bring up a good point....
It is quite possible that the majority of the population is no longer capable of having intellectual conversations at all.
I'll wager that a large portion of chat on the internet is orchestrated by what people saw in the media. What amazes me is how quickly topics of "great importance” suddenly fade into obscurity overnight and are replaced by something else equally irrelevant in a never ending cycle.
It is as if peoples brains are remote controlled by whatever the media owners decide that they should be thinking and talking about or doing.

It was bad enough when there was only televison and radio. At least back then there were moments however short where they were away from the media and not being brainwashed. Now in effect the television can follow them wherever they go.

The way people think has changed from what it once was. People no longer rely upon personal experience or information from trustworthy people they know, but often believe whatever they hear in the media from whatever talking head that they consider an authority figure.

The results of mass media have been disastrous not only for causing violent riots and wars , but more insidiously for lulling people into inaction, learned helplessness and being kidalts who only care about instant gratification.

I will revise what i said earlier as follows...
We need to get together and talk about things we care about with live people as well as continually think about what makes these things important to us. It takes a lot of extra effort to overcome the natural human tendency to mindlessly imitate others but it’s important.

I've done a lot of soul searching and I'm pretty sure i am genuinely interested in talking about pascal. My interest in pascal came from a pascal class I took long ago and has been enforced by years of usage. But there is always more to discover.

Would anyone here like to chat in real time about pascal too?

[MarkMLl]

I'm unhappy perpetrating a thread which is wandering off topic, but...

Personally, I believe the root of the problem is the severe lack of analytical skills in a very significant percentage of the population. Combine that with the desire (and sometimes demand) for instant gratification and there is the making of a severe and chronic problem.  Unfortunately, A.I is more likely to worsen this situation than improve it.  From what I've seen so far, A.I promotes intellectual inactivity (A.I will figure it for you...)

And regrettably, that applies to a lot of people who in principle have benefited from the educational system and should be able to do better.

However the biggest problem appears to be that the "media barons" now have a tool which allows them to tailor every page to the individual reader, and while we are used to the fact that advertising content is dynamic having content similarly variable is relatively new.

So while two people might think that they are reading the same material (the URL or TV channel number is the same), the right-winger sees survivalist adverts and reporting and comments cherry-picked to make the left look bad, and the left-winger sees advice on gender fluidity and reporting and comments cherry-picked to make the right look bad.

AI, as understood by the current generation, is inherently neither good nor bad. Neither is it intelligence: it's basically just pattern matching with access to what is by now an unimaginably large training corpus. Neither is it controllable, because while people like Geoffrey Hinton obviously know how it's put together they don't, in general, know how it works.

Frankly, I think we're screwed. There's nothing we can do to make things better, but we can try to avoid making things worse by not getting involved.

MarkMLl

[Joanna from IRC]

.  So while two people might think that they are reading the same material (the URL or TV channel number is the same), the right-winger sees survivalist adverts and reporting and comments cherry-picked to make the left look bad, and the left-winger sees advice on gender fluidity and reporting and comments cherry-picked to make the right look bad.

That’s part of the divide and conquer strategy to split us all up into a bunch of squabbling groups with unsolvable differences. What most people don’t realize however is that all the variety of opinion is just an illusion. The same interests are behind the backing of all sorts of Superficially unrelated   things.

Even if you refuse to get involved in whatever the latest version of distraction and hysteria is, it’s guaranteed that most others will not be so wise and plenty of resources will be squandered upon it. 

So far as I can determine, the craziness will not stop until all the Conditions which facilitate it are no longer possible.