Using chatgpt to write code? Yes!

[dbanner]

It can write code, you just have to outline some parameters, and in a matter of few minutes, it done. Super powerful. Super dangerous too!

[wtrk]

İs there anything to show us bro ?

[dbanner]

I saw a demonstration on YouTube of this guy using it to write some code for programming an Arduino to make a seven led chaser.
I messed around with it not for code but just to ask it some questions.
The replies I got was astonishing. It really got me scared. I stayed away after that. But for writing code, yeah, it's pretty darn impressive.

[cristiano.sar]

It works pretty good, I requested him to write me a code to make a web scraping script (capture information from a website) in a programming language I know and it made it very good, you can see it here:

[Teleno]

No wonder. It's ChatGPT's mother tongue!

[kosacid]

its good for ideas to base your code on but as for righting code it did get a few things wrong eg it gave me the timer addresses for a mega not a nano

[StevenRales1]

I need some guidance to write code on ChatGPT. I am working on a project. Tell me how can i get information about the right code.

[MartinB]

Just asked it to write code for arduino to monitor battery voltage and it came up with the correct code. I can only assume that it scanned the net to find the answer, but then again perhaps it actually wrote the code. Spooky !!!!

[MartinB]

Not that good. It couldn't write code for an arduino based metal detector.

[ivconic]

I don't know about the other, but as far as writing code for the Arduino; ChatGPT at OpenAI is still far from being a real "programmer".
I found out about the site relatively early on and immediately went to see what it was all about.
There was no end to my initial delight. Just the first 20 minutes...
Very simple things, like "Hello world!" will write correctly, almost without error.
But anything more advanced, closer to hardware; will write with many mistakes, often will write a bunch of gibberish.
As an Arduino programmer who mastered the basics a long time ago; I realized that it is extremely useless to me.
It makes incredible mistakes. Repeats mistakes. It often falls into dubious mistakes that pile up.
E.g; often something related to PWM will connect to a non-PWM pin or vice versa.
When it is given a complex task with multiple explanations; as a rule, it misses something from those explanations/demands to do.
When I remind it of what was missed; it does it and misses something else it did before. In the same code.
After a couple of days of playing I just gave up. It is still far from true AI. It is only good at the level of children's stories.
For more serious tasks it is absolutely useless and will generate tons of errors.
It is good enough for very simple tasks and as such is good for beginners who are just learning programming.
For programming that is not directly related to specific hardware, it is slightly better.
So that's why I came to the conclusion that it is usable when writing portable code.
But it should be done in short sequences. Because with more complicated structures, it is very easy to get confused and start making cumulative mistakes.
Conclusion; it's still a long way from true AI... but a good try for now. An interesting toy.

[MartinB]

The biggest problem is that people will believe the misinformation it provides.

[ivconic]

Yes, it will be a problem, but only when it comes to people without formal education and knowledge.
(Unfortunately; there are more and more of them).
That is why it is very important to maintain the system of formal education and improve it in the parts that are not good.
What I'm going to write next falls into the realm of conspiracy theories, but read it, think about it; and draw your own conclusions.
I am 56/7 years old and I remember what the attitude towards formal education was like xx years ago.
Not only the attitude of the common man, but the attitude of society, the state and the state system.
Nowadays, all that has changed.
There are more and more formally uneducated people.
And society and the state seem to be deliberately working to make them more.
The emergence of mass media and the Internet as the most powerful medium among the existing ones; has only deepened this problem.
Today, regular, formal education is easily replaced with instant internet "education".
Parents who do not have money to invest in their children's education; they look at this phenomenon as an alternative solution, even welcome!
And so millions of charlatans are created.
Today, everyone "knows everything". It is enough to briefly "google" what interests him and voila; he immediately "found out" and "learned".
Today, everyone is a "doctor", "philosopher", "politician", "scientist", "expert in this and that"...
Either all this is a combination of very unfortunate circumstances... or all this was thought up by someone very vile and very intelligent!
One of the conspiracy theories says that the goal of a certain privileged group of "shadow rulers" is to dumb down the largest number of the population on the planet.
There are several main conspiracy theories. But they all have the same goal. Keeping that group in power at any cost.
The rule of that group over the rest of the world. Population reduction. Etc.. you've probably heard of all those theories yourself.
Whether these are all false and stupid conspiracy theories... or the truth; the fact is that some results are already easily visible.
There are also totally idiotic conspiracy theories, like "flat earth", "no, they didn't land on the moon", "HAARP radiates and causes earthquakes",
"chemical traces from airplanes are poisoning us"....
Those very stupid and obviously incorrect theories were deliberately pushed, in order to completely discredit some true doubts and researches from the open minded people.
Well, one of the elements is that AI, which will supposedly replace humans in many ways.
No, it won't. Not in the near future.
And if one tries; it is clear that this will produce even more disasters.
For now, this is all a "trial bubble".
And of course, when it's already there; a means for additional matching of people from regular and formal education.
Yes, it is true, many young and inexperienced will believe. They are the target group of this experiment.

[MartinB]

Well my friend, we are two of a kind here. I appreciate this is not a political forum so I will not say much, but if we lived closer we could go for a beer and put the world to rights.

[ivconic]

What you said reminded me of how few friends I have today!! Everyone else is online.
And I only have one or two friends that I see in real life.
I had them more, long before the "online" era.
But life scattered us in all directions and separated us.
At home, the situation is the same.
Younger son in his room, wife in another room, me in the third... and we correspond via Viber, when we agree on things related to household duties, etc.
The older son is far away, he has been living in NY for several years and the only possible contact with him is via Viber and email.
My wife from the other room often writes to me on Viber: "Make coffee and when it's ready call me to come over" !!!
And often I write to her via Viber: "Come set lunch".
The only thing that connects us besides Viber is an incredibly restless cat that runs from room to room all day long.
This is where "evolution" has brought us...
Which tells me that AI will become the talk pal of many soon!


Regarding the previous post...
At first I believed that there were very few people who understood all this.
Fortunately, in frequent contact with people, I learned that there are many more.
But we don't all live in the same conditions. And not everyone is ready to speak publicly and honestly say what they think.
Because they are afraid that someone will "mark" them, that they will have problems because of it.
That's why most are silent. But that's not a sign that they don't know.
...
As for the AI, I noticed something else. Can write fairly decent C++ code for "general" stuff. Not tied to specific hardware.
So it can help real developers as an aid in writing more voluminous code and shorten their time.
But I underline; "real" developers, not beginners.
Because a beginner will not be able to debug the same code successfully and quickly.
And the loss of time in debugging then cancels out the gain in time in the use of AI.
And as for writing code for specific hardware, embedded devices, etc. that's where AI fails well almost always.
For writing various texts on various topics... the matter becomes comical, because the AI writes a bunch of unnecessary and boring sentences,
so it resembles some sick scribbler!
Everything I wrote applies to the free version. Since MS stepped in and took over AI, a new, soon-to-be-released, commercial version was created.
I have no experience with it.
The old story repeats itself, the free version was a "teaser" to "catch" as many customers as possible.
And in the end, everything revolves around money again.

[moodz]

... I dont like the "big cloud" AI models They can see every question you ask and its conceivable they ( the AIs ) can figure out what you are working on by the types of questions you ask.
I was able to download my own private AI neural net so and get it running ( though it insists on calling me Tim in chat mode .. when my name is Paul.
But it can generate code and it thinks Donald Trump is president and importantly it can be trained.




image widget