AI usefulness discussion

[440bx]

@440bx have you ever seen any opensource ai projects?

The one I am aware of is by a forum member: schuler.

His thread about it is: https://forum.lazarus.freepascal.org/index.php/topic,39049.0.html

[Joanna from IRC]

Wow I’m actually allowed to look at code on source forge 

I guess open source ai isn’t as bad as black box ai whose inner workings are unknown.

[regs]

@440bx have you ever seen any opensource ai projects?

Ollama has a number of models specifically for coding. Like codellama, qwen2.5-coder, deepseek-coder-v2, starcoder2. There are also general models that can do some coding, like llama, qwen2.5, deepseek-r1, phi4 etc etc.

I don't know how OpenAI and Copilot are doing, as they are only available in USA and just few more western countries, but most open models aren't good with Pascal. They can do some basics, but when you start digging into LCL, CCR etc non can produce compilable code, nor even know much of those libraries.

Even more, codellama explicitly stating that https://sourceforge.net/p/lazarus-ccr/svn/ is not an official repository and even if, it cannot discuss it "without proper authorization and consent from the developers and community members who contribute to this project".

[TRon]

None of the models are opensource. Do they explicitly state how the model was trained and on what data ?

I do not give one crappers about how f.e. the EU defines open source for AI because it is bullocks (they should be hung for redefining the definition of open source).

fwiw we have had our share of AI scraping, ask the admin how they ignored the settings. No-one ever asked consent and 'they' just started scraping. That one of them states to not have done so is admirable but I wouldn't trust any of those statements.

All imho ofc.

[Joanna from IRC]

I am not a programmer, I don't know how to code, but using ChaTGPT I was able to simply create a carousel component to display images with the help of ChaTGPT

I don’t understand why these people who are not interested in learning how to program in pascal keep coming here to shove chat gpt down our throats. Isn’t there a chat gpt forum somewhere? 

[cdbc]

Hi
Hahaha Joanna, just do what I do  ...and look the other way 
Regards Benny

[Thaddy]

BTW, Joanna,
Schuler's more recent code can run some LLM's!! Locally.
Takes quite some hardware to be of use, but it works.

Then again, he got his Ph.d in computer science because of that in 2022.
Using Freepascal ...
Joao is quite well known, not only here.
https://www.researchgate.net/profile/Joao-Paulo-Schwarz-Schuler

[Joanna from IRC]

If ai is so “smart” can it solve this? Or similar things? That it hasn’t scraped off the Internet? 

There are five houses.
The Englishman lives in the red house.
The Spaniard owns the dog.
Coffee is drunk in the green house.
The Ukrainian drinks tea.
The green house is immediately to the right of the ivory house.
The Old Gold smoker owns snails.
Kools are smoked in the yellow house.
Milk is drunk in the middle house.
The Norwegian lives in the first house.
The man who smokes Chesterfields lives in the house next to the man with the fox.
Kools are smoked in the house next to the house where the horse is kept.
The Lucky Strike smoker drinks orange juice.
The Japanese smokes Parliaments.
The Norwegian lives next to the blue house.
Now, who drinks water? Who owns the zebra?

[kapibara]

If ai is so “smart” can it solve this? Or similar things? That it hasn’t scraped off the Internet? 

There are five houses.
The Englishman lives in the red house.
The Spaniard owns the dog.
Coffee is drunk in the green house.
The Ukrainian drinks tea.
The green house is immediately to the right of the ivory house.
The Old Gold smoker owns snails.
Kools are smoked in the yellow house.
Milk is drunk in the middle house.
The Norwegian lives in the first house.
The man who smokes Chesterfields lives in the house next to the man with the fox.
Kools are smoked in the house next to the house where the horse is kept.
The Lucky Strike smoker drinks orange juice.
The Japanese smokes Parliaments.
The Norwegian lives next to the blue house.
Now, who drinks water? Who owns the zebra?

AI: This puzzle is a variation of the famous "Einstein's riddle." We'll use the provided clues to determine who drinks water and who owns the zebra.

Let's break down the information systematically:

    House Layout: There are five houses in a row, numbered from 1 to 5.
    Clues:
        (1) The Englishman lives in the red house.
        (2) The Spaniard owns the dog.
        (3) Coffee is drunk in the green house.
        (4) The Ukrainian drinks tea.
        (5) The green house is immediately to the right of the ivory house.
        (6) The Old Gold smoker owns snails.
        (7) Kools are smoked in the yellow house.
        ( Milk is drunk in the middle house (House 3).
        (9) The Norwegian lives in the first house.
        (10) The man who smokes Chesterfields lives in the house next to the man with the fox.
        (11) Kools are smoked in the house next to the house where the horse is kept.
        (12) The Lucky Strike smoker drinks orange juice.
        (13) The Japanese smokes Parliaments.
        (14) The Norwegian lives next to the blue house.

Now, let’s fill in the information step-by-step based on the clues.
Step 1: Determining basic information from the clues

    House 1: Norwegian (Clue 9)
    House 3: Milk (Clue

Since House 3 drinks milk, it cannot be green (since coffee is in the green house). Therefore, based on Clue 5, House 3 must be ivory and House 4 must be green (so coffee is there).
Current Status:

 
House 1: Norwegian, ?, ?, ?, ?
House 2: ?, ?, ?, ?, ?
House 3: ?, Ivory, Milk, ?, ?
House 4: ?, Green, Coffee, ?, ?
House 5: ?, ?, ?, ?, ?

Step 2: Continuing to fill in based on clues

    From Clue 14, if House 1 is the Norwegian's house, House 2 must be blue.
    From Clue 1, the Englishman must be in the red house, which could only be House 5 since House 3 is ivory and House 4 is green (and House 1 is yellow).

Current Status:

 
House 1: Norwegian, Yellow, ?, ?, ?
House 2: ?, Blue, ?, ?, ?
House 3: ?, Ivory, Milk, ?, ?
House 4: ?, Green, Coffee, ?, ?
House 5: Englishman, Red, ?, ?, ?

Step 3: Fill in the remaining information

    From Clue 7, Kools are in House 1 (the yellow house).
    From Clue 4, the Ukrainian drinks tea, which means he must be in House 2.
    The Lucky Strike smoker drinks orange juice (Clue 12), so he must be in House 5.
    From Clue 6, the Old Gold smoker owns snails, which leaves House 4 for him, and thus House 5 must be the Spaniard (who owns the dog).

Current Status:

 
House 1: Norwegian, Yellow, Kools, ?, ?
House 2: Ukrainian, Blue, Tea, Horse, Chesterfields
House 3: ?, Ivory, Milk, ?, ?
House 4: Old Gold, Green, Coffee, Snails, ?
House 5: Spaniard, Red, Orange Juice, Dog, ?

Final Fill-In

We can deduce:

    The Japanese must be in House 3 because he smokes Parliaments (Clue 13).
    The only pet left is the zebra, who must belong to the Japanese since all others are accounted for.

Conclusion:

    The person who drinks water is the Norwegian in House 1 (because all other drinks are accounted for).
    The Japanese owns the zebra.

So:

    Who drinks water? The Norwegian.
    Who owns the zebra? The Japanese.

[Joanna from IRC]

I probably should have picked something more original . That doesn’t have solution on internet already 
Actually that has alt of errors in it using ai.
I know because I solved it I’m not sure how it arrived at correct answers at all.

[regs]

None of the models are opensource.

Some of them are. But the main point is that they are all private for offline use.

A new Grok 3 isn't open and isn't for self-host, but showing very strong hope for Pascal. It knows it very well.

[TRon]

Some of them are.

None of the major players only perhaps maybe some small projects (that have no significance). Feel free to enlighten me/us.

But the main point is that they are all private for offline use.

Private is not the same as open source. Those are two completely different topics.

I am in complete disagreement with the official EU redefinition of open-source. Open-source means everything is open, including training parts/data. And that is besides the bias of the data itself and artificial restrictions.

A new Grok 3 isn't open and isn't for self-host, but showing very strong hope for Pascal. It knows it very well.

I have no opinion on grok because I've never used it.