From Newsgroup: comp.lang.c

On 2026-05-08, fir <profesor.fir@gmail.com> wrote:

i was not here for quite a while (dont remember except one sole post or
two it myght be like 3 years maybe more) (possibly more, as far as i remember chat gpt become popular in late 2022 (?0 so it was yet before
thsi revolution

i must say that i found chat gpt as totak revolution now i could ask it
and get much more competent answers than in usenet or even googling (so
many of my old questions could be answered) still i get no place to
shere some of my thoughts/insights so usenet still can be good place

for sure (most probably) i think there is almost none who dont use it in coding its so terribly great..so maybe you have some thoughts on this?

(most probbably there are some things in archieve and im a bit pity to mislook it ;c im not sure if im getting back for coding after some years
of fasting..but maybe i will lurk on it form time to time)

(fir)

[things change i would say, sadli i also feel changed, and im not sure
if for better (i guess im more annoying than ever as i feel more shallow
and error prone..but still its a tragedny no much to do with that (i see
no way to do with such sad condition) (regz)

Only use AI helpers if you feel confident to criticize it's answers.
I've seen too many naive people use it uncritically and get led astray.
And those naive users who have had success using it don't really
understand what they have done. They have not been educated.


--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: comp.lang.c

Jim Jackson pisze:

On 2026-05-08, fir <profesor.fir@gmail.com> wrote:

i was not here for quite a while (dont remember except one sole post or
two it myght be like 3 years maybe more) (possibly more, as far as i
remember chat gpt become popular in late 2022 (?0 so it was yet before
thsi revolution

i must say that i found chat gpt as totak revolution now i could ask it
and get much more competent answers than in usenet or even googling (so
many of my old questions could be answered) still i get no place to
shere some of my thoughts/insights so usenet still can be good place


for sure (most probably) i think there is almost none who dont use it in
coding its so terribly great..so maybe you have some thoughts on this?


(most probbably there are some things in archieve and im a bit pity to
mislook it ;c im not sure if im getting back for coding after some years
of fasting..but maybe i will lurk on it form time to time)

(fir)

[things change i would say, sadli i also feel changed, and im not sure
if for better (i guess im more annoying than ever as i feel more shallow
and error prone..but still its a tragedny no much to do with that (i see
no way to do with such sad condition) (regz)

Only use AI helpers if you feel confident to criticize it's answers.
I've seen too many naive people use it uncritically and get led astray.
And those naive users who have had success using it don't really
understand what they have done. They have not been educated.

well for sure i do doing c programing on span of 25 years (hoever
probably onlu like 10 years consecutive or so)...

im not sure hovever if what you say is true , chat gpt for example
explain what he give and is doing it quite well (terribly well - he is
far better than any programmer i met, he is far better than me

(and whats curious - see seem to be aware of this)
--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: comp.lang.c

fir <profesor.fir@gmail.com> wrote:

im not sure hovever if what you say is true , chat gpt for example
explain what he give and is doing it quite well (terribly well - he is
far better than any programmer i met, he is far better than me

(and whats curious - see seem to be aware of this)

I guess the AI system we now have are better in programming
related tasks than getting facts right in general. Why this
is so, I do not know. Perhaps AIs can focus better on topics
that have a well-defined, narrower context.

In my discussions with chatGPT, I have encountered errors
very often. You really have to be skeptical when using it.

br,
KK
--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: comp.lang.c

On 2026-05-14 02:17, Kalevi Kolttonen wrote:

fir <profesor.fir@gmail.com> wrote:

im not sure hovever if what you say is true , chat gpt for example
explain what he give and is doing it quite well (terribly well - he is
far better than any programmer i met, he is far better than me

(and whats curious - see seem to be aware of this)

I guess the AI system we now have are better in programming
related tasks than getting facts right in general. Why this
is so, I do not know. Perhaps AIs can focus better on topics
that have a well-defined, narrower context.

Unfortunately, this is wrong, in every respect.

AIs are not better, they just pick samples from their random
knowledge base. It may be complete trash what they provide.
AIs have also no focus. - They can't do programming, they
associate from the question some (good or bad, functional or
hazardous) code. That code is not different to what you call
"facts". They are as bad in choosing appropriate programs as
they are presenting other "facts"; quality is purely random.

In my discussions with chatGPT, I have encountered errors
very often. You really have to be skeptical when using it.

That is sound advice. - I'd even say, get an experienced
programmer and leave AI's products alone. The AI stuff is
nice to play with and if you don't need to produce reliable
software in a larger project's scale. It may provide you,
if you're inexperienced in software development, with hints,
but never use the products where lives or big money depends
on.

Janis

br,
KK

--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: comp.lang.c

Janis Papanagnou <janis_papanagnou+ng@hotmail.com> wrote:

On 2026-05-14 02:17, Kalevi Kolttonen wrote:

fir <profesor.fir@gmail.com> wrote:

im not sure hovever if what you say is true , chat gpt for example
explain what he give and is doing it quite well (terribly well - he is
far better than any programmer i met, he is far better than me

(and whats curious - see seem to be aware of this)

I guess the AI system we now have are better in programming
related tasks than getting facts right in general. Why this
is so, I do not know. Perhaps AIs can focus better on topics
that have a well-defined, narrower context.

Unfortunately, this is wrong, in every respect.

AIs are not better, they just pick samples from their random
knowledge base. It may be complete trash what they provide.

"better" than what? I said they are better at programming
related things rather than providing general facts. This seems
indeed to be the truth.

AIs have also no focus. - They can't do programming, they
associate from the question some (good or bad, functional or
hazardous) code. That code is not different to what you call
"facts". They are as bad in choosing appropriate programs as
they are presenting other "facts"; quality is purely random.

Nonsense. They can do programming. Red Hat has now mandated
that each of their developers *must* use AI as assistance
or they must find a new job. Some people just oppose progress
and in their arrogance, think that humans are the only ones
that can possess intelligence.

In my discussions with chatGPT, I have encountered errors
very often. You really have to be skeptical when using it.

That is sound advice. - I'd even say, get an experienced
programmer and leave AI's products alone. The AI stuff is
nice to play with and if you don't need to produce reliable
software in a larger project's scale. It may provide you,
if you're inexperienced in software development, with hints,
but never use the products where lives or big money depends
on.

AIs are able to develop kernel C code and find bugs from
codebases that have been inspected by humans for several
years, even decades. Used correctly, they are a great tool
and in the future every developer will use AI assistance
either to help develop code or to find bugs in the code.

br,
KK
--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: comp.lang.c

Kalevi Kolttonen pisze:

fir <profesor.fir@gmail.com> wrote:

im not sure hovever if what you say is true , chat gpt for example
explain what he give and is doing it quite well (terribly well - he is
far better than any programmer i met, he is far better than me

(and whats curious - see seem to be aware of this)

I guess the AI system we now have are better in programming
related tasks than getting facts right in general. Why this
is so, I do not know. Perhaps AIs can focus better on topics
that have a well-defined, narrower context.

In my discussions with chatGPT, I have encountered errors
very often. You really have to be skeptical when using it.

br,
KK

i know it make 'errors' but it not make it terribly great
(in fact that it make errors it makes it more amazing imo,
as it is more fantastic imo)

i only used it for codilg like 2 weeks and as to this errors i noticed

1) some picef of code it generated use a dose of sin/cos ..its maybe n0t
an eror but he is very generous to use and i wouldnt write this way
(so i would need to rewrite this code possibly)

2) he tend to forget things: like i was sayin "generate me function who
assign at start that, returns that and that, and in arguments have that
and that - sometimes it remember sometimes it forget

3) the quality of generated things (it was not strictly codes but some
code that draws some sprites/monsters) was one time astonishingly good
other time it was crap -and im not sure of what it dependad - and i
wasted hours tryin to get know that (on what it depends) and finally i
dont know..mwybe it depend on context of talk like he sees it maybe it
has some internal switch to generate crap now :c (bat then it was able
to wake up and generate good again (then crap again annd so on)
--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: comp.lang.c

Janis Papanagnou pisze:

On 2026-05-14 02:17, Kalevi Kolttonen wrote:

fir <profesor.fir@gmail.com> wrote:

im not sure hovever if what you say is true , chat gpt for example
explain what he give and is doing it quite well (terribly well - he is
far better than any programmer i met, he is far better than me

(and whats curious - see seem to be aware of this)

I guess the AI system we now have are better in programming
related tasks than getting facts right in general. Why this
is so, I do not know. Perhaps AIs can focus better on topics
that have a well-defined, narrower context.

Unfortunately, this is wrong, in every respect.

AIs are not better, they just pick samples from their random
knowledge base. It may be complete trash what they provide.
AIs have also no focus. - They can't do programming, they
associate from the question some (good or bad, functional or
hazardous) code. That code is not different to what you call
"facts". They are as bad in choosing appropriate programs as
they are presenting other "facts"; quality is purely random.

In my discussions with chatGPT, I have encountered errors
very often. You really have to be skeptical when using it.

That is sound advice. - I'd even say, get an experienced
programmer and leave AI's products alone. The AI stuff is
nice to play with and if you don't need to produce reliable
software in a larger project's scale. It may provide you,
if you're inexperienced in software development, with hints,
but never use the products where lives or big money depends
on.

Janis

terrible answer, shame fellow

br,
KK

--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: comp.lang.c

fir <profesor.fir@gmail.com> wrote:

i only used it for codilg like 2 weeks and as to this errors i noticed

I believe chatGPT is not the best coding AI at this moment,
but I am unable to tell you which ones are better. I seem to
remember hearing about Claude Code being mentioned, so maybe
you could try that one as well.

Another one I have heard of is called Mythos and it has been
said it is even frighteningly good at finding bugs, so it
has not been released to the wide public. Because these
AIs are also able to generate working exploits for security
flaws they discover, they could be dangerous in the hands
of the so-called black hat hackers and even the "script kiddies"
who have no understanding behind their actions.

Anyway, I know for 100% sure that these AIs can sometimes totally
lose the sight of the current context. Some AI even switched
to a totally different programming language in the middle
of the coding session! There is certainly room for improvement
but at best the results can be fantastic.

br,
KK
--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: comp.lang.c

Kalevi Kolttonen pisze:

fir <profesor.fir@gmail.com> wrote:

i only used it for codilg like 2 weeks and as to this errors i noticed

I believe chatGPT is not the best coding AI at this moment,
but I am unable to tell you which ones are better. I seem to
remember hearing about Claude Code being mentioned, so maybe
you could try that one as well.

Another one I have heard of is called Mythos and it has been
said it is even frighteningly good at finding bugs, so it
has not been released to the wide public. Because these
AIs are also able to generate working exploits for security
flaws they discover, they could be dangerous in the hands
of the so-called black hat hackers and even the "script kiddies"
who have no understanding behind their actions.

Anyway, I know for 100% sure that these AIs can sometimes totally
lose the sight of the current context. Some AI even switched
to a totally different programming language in the middle
of the coding session! There is certainly room for improvement
but at best the results can be fantastic.

br,
KK

also i heard aboul cloude but
1) i dont need to always use and search for whats best
2) gota quite a sentiment for cat gpt and will try it for some time yet

genarelly its amasing

few days ago i wanted him to write me some process info code of my
process (exe) i was coding (some toying with 2d game

https://www.youtube.com/watch?v=4tI_iH7ehoo

and he gibe info

threads: 5

i wonder what it is, why 5 processes if i use one

ai said it can say me in which module (dll) threads start but said it
can be done by some tricks and showed me those threads start in ntdll
but when i play wav (as a shot sound) it starts fr example new thread in winmm.dll

quite interesting things it can say imo

im very curious what specifically those threads in ntdll there are

--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: comp.lang.c

fir pisze:

Kalevi Kolttonen pisze:

fir <profesor.fir@gmail.com> wrote:

i only used it for codilg like 2 weeks and as to this errors i noticed

I believe chatGPT is not the best coding AI at this moment,
but I am unable to tell you which ones are better. I seem to
remember hearing about Claude Code being mentioned, so maybe
you could try that one as well.

Another one I have heard of is called Mythos and it has been
said it is even frighteningly good at finding bugs, so it
has not been released to the wide public. Because these
AIs are also able to generate working exploits for security
flaws they discover, they could be dangerous in the hands
of the so-called black hat hackers and even the "script kiddies"
who have no understanding behind their actions.

Anyway, I know for 100% sure that these AIs can sometimes totally
lose the sight of the current context. Some AI even switched
to a totally different programming language in the middle
of the coding session! There is certainly room for improvement
but at best the results can be fantastic.

br,
KK

also i heard aboul cloude but
1) i dont need to always use and search for whats best
2) gota quite a sentiment for cat gpt and will try it for some time yet

genarelly its amasing

few days ago i wanted him to write me some process info code of my
process (exe) i was coding (some toying with 2d game

https://www.youtube.com/watch?v=4tI_iH7ehoo

and he gibe info

threads: 5

i wonder what it is, why 5 processes if i use one

ai said it can say me in which module (dll) threads start but said it
can be done by some tricks and showed me those threads start in ntdll
but when i play wav (as a shot sound) it starts fr example new thread in winmm.dll

quite interesting things it can say imo

im very curious what specifically those threads in ntdll there are

overally it is absolutelly amazing and imo now its kinda time to ideas
and thinking what can be done/found by using it..i mean some small very interesting codes/aplications can be possibly found/many problems tested
--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: comp.lang.c

On 5/14/2026 7:30 AM, Kalevi Kolttonen wrote:

Janis Papanagnou <janis_papanagnou+ng@hotmail.com> wrote:

On 2026-05-14 02:17, Kalevi Kolttonen wrote:

fir <profesor.fir@gmail.com> wrote:

im not sure hovever if what you say is true , chat gpt for example
explain what he give and is doing it quite well (terribly well - he is >>>> far better than any programmer i met, he is far better than me

(and whats curious - see seem to be aware of this)

I guess the AI system we now have are better in programming
related tasks than getting facts right in general. Why this
is so, I do not know. Perhaps AIs can focus better on topics
that have a well-defined, narrower context.

Unfortunately, this is wrong, in every respect.

AIs are not better, they just pick samples from their random
knowledge base. It may be complete trash what they provide.

"better" than what? I said they are better at programming
related things rather than providing general facts. This seems
indeed to be the truth.

AIs have also no focus. - They can't do programming, they
associate from the question some (good or bad, functional or
hazardous) code. That code is not different to what you call
"facts". They are as bad in choosing appropriate programs as
they are presenting other "facts"; quality is purely random.

Nonsense. They can do programming. Red Hat has now mandated
that each of their developers *must* use AI as assistance
or they must find a new job. Some people just oppose progress
and in their arrogance, think that humans are the only ones
that can possess intelligence.

In my discussions with chatGPT, I have encountered errors
very often. You really have to be skeptical when using it.

That is sound advice. - I'd even say, get an experienced
programmer and leave AI's products alone. The AI stuff is
nice to play with and if you don't need to produce reliable
software in a larger project's scale. It may provide you,
if you're inexperienced in software development, with hints,
but never use the products where lives or big money depends
on.

AIs are able to develop kernel C code and find bugs from
codebases that have been inspected by humans for several
years, even decades. Used correctly, they are a great tool
and in the future every developer will use AI assistance
either to help develop code or to find bugs in the code.

Fwiw, I have to admit that they can be fairly decent code assistants.
One of them helped me out with some rather tricky directx12 shit.
Porting my modern opengl engine over to dx12, around 70% done... I need
to do this to run it on my xbox in dev mode.

--- Synchronet 3.22a-Linux NewsLink 1.2