Interesting article about AI

[lainz]

I think a lot of people don’t understand what programming is 

From me is making instructions to a machine. Translating from human language to machine code.

That can be done with assembler, with Pascal or with AI. All do the same.

[Zvoni]

I’ve often found that in order to ask good questions you have to be knowledgeable about the topic that you’re asking questions about or even know what the correct answer is already.

Disagree.
One of the most important questions is "Why?"
And you don't have to know anything about the topic, but the one you question SHOULD know what he/she's talking about, since you are asking for the Reason.
And how many of those "smart" people fail to answer such a simple question as "Why?"

I think a lot of people don’t understand what programming is 

Starting school at 5/6 years old.
From personal experience: I started school at 6 years old in Germany. Learning my letters and numbers....
then 2 years later, my parents decided, that i had to attend an "additional" school.
Yeah....right....after 2 years learning the western Alphabet, trying to learn the cyrillic alphabet at 8 years old was like trying to learn klingon.

"Programming" starts as early as cognitive processes are established --> "Don't touch that hot pan", "Don't stick the knitting needle into a power outlet"

Pavlov's Dogs, anybody?

And FWIW regarding the whole AI-Discussion:
An AI will never create a 9th Symphony, never paint a Mona Lisa.
The Moment an AI actually acomplishes something like that, i'll be the first in line demanding to pull the plug

[440bx]

Why do you think it’s so important to just strive for “productivity” at all costs.

"at all costs" is likely a bad idea, however striving for productivity is a _necessity_.  There are over 7 billion people, feeding and clothing (among other things) that many people could not be done with the productivity of a hundred years ago.

What price are we willing to pay for increased productivity? Would it be ok to eliminate all superfluous members of the population? That would certainly increase productivity. What benefits does “increased productivity” is it really necessary to replace all possible jobs with substandard bots.

Go to a mirror and ask yourself that.  You use OOP and the LCL in spite of the fact that the resulting program is _much_ larger, somewhat slower and much harder to maintain than the equivalent written without OOP and the LCL.  You do it because it makes you feel more productive and the LCL encapsulates knowledge you don't have.  Some of the stones you're throwing at A.I are shaped like boomerangs.

The problem is that often but, not always, productivity comes at the expense of quality, as artificial things usually do.

[lainz]

Why do you think it’s so important to just strive for “productivity” at all costs.

"at all costs" is likely a bad idea, however striving for productivity is a _necessity_.  There are over 7 billion people, feeding and clothing (among other things) that many people could not be done with the productivity of a hundred years ago.

What price are we willing to pay for increased productivity? Would it be ok to eliminate all superfluous members of the population? That would certainly increase productivity. What benefits does “increased productivity” is it really necessary to replace all possible jobs with substandard bots.

Go to a mirror and ask yourself that.  You use OOP and the LCL in spite of the fact that the resulting program is _much_ larger, somewhat slower and much harder to maintain than the equivalent written without OOP and the LCL.  You do it because it makes you feel more productive and the LCL encapsulates knowledge you don't have.  Some of the stones you're throwing at A.I are shaped like boomerangs.

The problem is that often but, not always, productivity comes at the expense of quality, as artificial things usually do.

I agree. But seems that the mainstream is like that. For example Python beign the first language In the Tiobe index. Like or not the index anyways is there.

[VisualLab]

Why do you think it’s so important to just strive for “productivity” at all costs.

"at all costs" is likely a bad idea, however striving for productivity is a _necessity_.  There are over 7 billion people, feeding and clothing (among other things) that many people could not be done with the productivity of a hundred years ago.

And it is this gradually increasing productivity in various fields that has resulted in the current 7 billion people. And they continue to grow: productivity and the number of people

You use OOP and the LCL in spite of the fact that the resulting program is _much_ larger, somewhat slower and much harder to maintain than the equivalent written without OOP and the LCL.  You do it because it makes you feel more productive and the LCL encapsulates knowledge you don't have.

Can code using OOP be larger and/or slower in some cases than code using only structures and subroutines – yes, maybe, sometimes. It probably depends on what specific objects we are using, what program we are creating, what problems it is solving. And yes, it is true that we use objects (LCL, etc.) because we do not always have the knowledge of how to solve a specific problem. And that they shorten the time of creating a program (i.e. the aforementioned: productivity). But that is what programming has always been about – reusing code once written, without reinventing the wheel.

In the case of procedural programming, we also often use procedures/functions that are placed in libraries (RTL, etc.). And we use them for the same reason as classes - because we do not always have knowledge about how to solve a specific problem. So library procedures/functions are used to solve a problem, but in a slightly different way than classes. The difference between OOP and a set of procedures/functions is a matter of scale - the detail of your own source code. Procedures/functions require:

• using specific data structures, as arguments to subroutines or results of function operation - an obvious thing,
• calling functions in a specific order (often: very strictly defined) - this is also obvious.

And this second point is just as important a reason for using OOP as productivity (and can affect it). When I have a ready object, I don't have to know so many technical details related to the order of calling specific subroutines (perhaps hundreds or thousands). Because the object already has implemented calls between individual subroutines (methods). Of course, you can't completely avoid calling subroutines, because objects have methods that also often require arguments. But the multitude of minutiae and technical details involved in procedure calls is significantly reduced. Of course, after some time, interested people who have access to the source code of classes will follow it and understand how specific classes work. But this needs time.

So OOP not only shortens program development time, but significantly reduces the amount of detail a programmer needs to know. And it's not just that such a programmer is lazy and ignorant. The same could be said about programmers who use libraries of procedures/functions - seemingly lazy and ignorant, because they don't write it themselves. Because it's not only knowledge (with a capital K) how to solve a given problem, but very often even details and technical details, such source code that has been polished for years in terms of performance and memory consumption.

And finally, there is the issue of this overhead on OOP. If you do not use classes, you have to write your own procedural source code (glue code, boilerplate), responsible for calling library subroutines, receiving results from functions, etc. Is this own code so minimal (thin) that it is almost negligible? Because I have doubts about it. Sometimes this code is "thin" and sometimes not. And without this code, you cannot write a useful program based on the structural-procedural paradigm. And classes simply contain part of such code. So what is this OOP overhead really like?

And one more thing, is every OOP the same, if we take into account the overhead of code size and the drop in performance due to the use of classes? Or maybe there are differences between OOP solutions? Because I claim that these differences in OOP are in individual groups of languages: from emulated OOP (Vala), through hybrid (C++, Object Pascal) to completely OOP (C#, Java). And on the fringes of OOP there are pseudo-OOP (JavaScript, Python). And this overhead is only to a small extent due to the presence of OOP itself, and depends more on the technical details, on how it was implemented.

In any case, classes should definitely be implemented in such a way that they are as simple as possible. And you shouldn't use OOP everywhere, without thinking (but how to do it in C# or Java?).

[Joanna from IRC]

I think a lot of people don’t understand what programming is 

From me is making instructions to a machine. Translating from human language to machine code.
That can be done with assembler, with Pascal or with AI. All do the same.

I’ll have to agree with zvoni... Programming was around long before computers. How about baking a cake. It is very complicated starting with being able to locate a kitchen that you are allowed to use and know how to identify and utilize everything in it properly. It’s also important to notice if anything is missing or unusable or dirty and remedy that problem before proceeding. To do this entails being able to think in terms of objects. I challenge 440bx to tell me how to bake a cake without using objects. 

zvoni I agree that people who genuinely understand what they are talking about should be able to explain it to a child. Maybe my lack of knowledge made me feel inadequate. It could have been that the people I was talking to were lacking the ability to teach me.

The fact that 7 billion people are alive on earth is not proof of productivity. The way they are being fed,clothed,housed transported and entertained is causing extreme environmental destruction and is only possible because of a temporary abundance of cheap fossil fuels. It’s a lot like one of those aquatic algae blooms that grow very well until they use up all the nutrients.

Productivity is the antithesis of quality. Productivity often tricks people into thinking that a lot is being accomplished when it isn’t.  Things look great on the outside but are shoddy and ready to break in the places you can’t see. This phenomenon is probably much worse with software than it is with poorly made clothing.

A factory that makes twice as many jackets by cutting corners is automatically considered to be twice as productive as one that makes jackets correctly so that they last a long time. Both jackets can look identical and be made of same materials but they are far from equal.

It’s not that people are incapable of making things properly. It’s that they are encouraged to create junk. They don’t even think about the consequences of the junk they create they just want money. This is why there is poorly sewn clothing with failing seams and airplanes with malfunctioning software that crashes them.

I think it’s about time to stop ignoring the big picture when it comes to productivity. AI will not bring utopia, it will only bring a worse version of what we already have.

[440bx]

Productivity is the antithesis of quality.

While that statement can be true the implication is completely fallacious.

Just an example, a reasonably well built car, responsibly driven and maintained will easily run for 100K+ miles.  Try getting that out of an early car.  I'd say good luck but, in this case that would be the stuff of miracles.

Bottom line: increases in productivity can also result in increases in quality.  Unfortunately that is not always the case.


I will not engage in a discussion about OOP but, I will re-state what I've already said too often: procedural programming can _always_ produce greater quality code than OOP. 

OOP is a religion.  People just want to believe in it, facts and logic are absent in that realm.

[LV]

The widespread integration of artificial intelligence into our lives is an undeniable reality. It is not influenced by my acceptance or rejection, admiration or resistance. However, having a solid understanding of your subject area is important when asking questions and using the answers provided by AI.

[Khrys]

And finally, there is the issue of this overhead on OOP. If you do not use classes, you have to write your own procedural source code (glue code, boilerplate), responsible for calling library subroutines, receiving results from functions, etc. Is this own code so minimal (thin) that it is almost negligible? Because I have doubts about it. Sometimes this code is "thin" and sometimes not. And without this code, you cannot write a useful program based on the structural-procedural paradigm. And classes simply contain part of such code. So what is this OOP overhead really like?

Heavy use of virtual functions (one of the cornerstones of OOP) doesn't play well with modern superscalar processors that use branch prediction. Branches with static targets are already bad enough (with people reporting 2.4x to 6x slowdowns in an example that misses 50% of branches), but indirect branches are even harder to predict since they have far more more than two possible outcomes (the first answer reports an overhead of 21 cycles on PPC, and Agner Fog quotes a misprediction penalty of 10-30 cycles on x86).

This of course only matters for very performance-critical code (and may in some cases be optimized away by the compiler, i.e. FPC's  DEVIRTCALLS), and in many cases having clean code outweighs having fast code.

[phoenix27]

It will come a point when human thinking will become negligible and AI will do all the thinking. That's the plan. Now they are just training it with the help of stupid humans who are too dumb to comprehend what's really happening. It's much more serious than the problem of losing a slave job. And what will happen whan AI will become really powerful? First they(who's they? Your OWNERS of course!) will put it as a layer between them and all of you. No more human judges, no more human police, no more human bureaucrats and so on and so forth. They can be too emotional and compassionate. And that's not good for your owners. Then, and that's the end goal, Ai will be inserted into the new generations upon birth. How, you may ask? With the so called "vaccines" that they started  experimenting on you all just some years ago. Did you forget what happened some years ago? Do you still think there was a pandemic? Ahahhahah!! That's not a conspiracy theory, that's a conspiracy FACT. At that point you will get your THOUGHTS(hence your indoctrination) directly from AI and you will finally think and act the way they want. No more indoctrination through school, TV and media in general, which works less and less, infact the term "fake news" has become widely spread, to the point that even the actor Trump uses it.

[Joanna from IRC]

Productivity is the antithesis of quality.

While that statement can be true the implication is completely fallacious.

Just an example, a reasonably well built car, responsibly driven and maintained will easily run for 100K+ miles.  Try getting that out of an early car.  I'd say good luck but, in this case that would be the stuff of miracles.

Bottom line: increases in productivity can also result in increases in quality.  Unfortunately that is not always the case.

I will not engage in a discussion about OOP but, I will re-state what I've already said too often: procedural programming can _always_ produce greater quality code than OOP. 

OOP is a religion.  People just want to believe in it, facts and logic are absent in that realm.

Efficiency can be very counterproductive here is an example https://youtu.be/iTaiLlQezNE

I am well aware of your hatred of oop and it inspires me to write better oop code 

[440bx]

I am well aware of your hatred of oop and it inspires me to write better oop code 

The only good OOP code is the OOP code that remains unwritten.  That's the OOP code that's really good.

[Joanna from IRC]

[quote author=440bx link=topic=69544.msg540600#msg540600 date=1733818626]

I am well aware of your hatred of oop and it inspires me to write better oop code 

The only good OOP code is the OOP code that remains unwritten.  That's the OOP code that's really good.
[/quote]

How about oop code that is deleted? 
I erase oop code and replace with fewer lines with better oop code..

I’ve asked you how to build a house and how to bake a cake without using oop. But you didn’t seize the opportunity to  tell me how to do things I understand without using objects. For one thing I’m curious how you control the access to data without any private or protected sections? Are you relying upon the interface and implementation sections to control access? That seems like it wouldn’t be as easy to do.

[Zvoni]

*snip* Now they are just training it with the help of stupid humans who are too dumb to comprehend what's really happening. *snip*

Exactly!
It doesn't matter, if it's an AI or a class full of children: If you only teach them chinese, why would anyone expect perfect english?

It's common in the industry: Shit in, shit out.
There is no Philosopher's Stone, which can make gold out of crap

[Thaddy]

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.

Well, well, well: do you trust the Encyclopedia's that your parents had from theirs, or bought for you for your studies?
I have Elsevier, Spectrum, Brittanica and a lithuanian one. Noone of them are currently reliable.