Artificial intelligence.

[Fernando Perez]

A little philosophical opinion.
Yesterday I needed two formulas to calculate two number sequences for a program. (see [Local Link Removed for Guests]).
I was very good at that not so many years ago. But I was totally stuck. To test, I went to ChatGPT OpenAI.
Within seconds, I had my formulas up and running:

Code: [Local Link Removed for Guests]

SUB displayIcon(icon, x, y)
  xi = (icon MOD 5) * 100
  yi = 200 - (int((icon/5)) * 100)
  vgaout.sprite x, y, w, h, xi, yi
  vgaout.show
END SUB

Surprise, satisfaction and .... My mental Meerkats woke up agitated.
How long ago, when I obtained my first square root on my first calculator, feeling that in a few years I would no longer know how to do it myself. Or that the tables of logarithms, paper edition, would disappear.
But it's so comfortable!

image.png

[PeterN]

Hello Fernando,
Yes, I see it similarly. There you have a tool whose impact on our hobby can already be enormous, fortunately depending on whether you actually want it or not. But the impact on e.g. the commercial area of software development is already enormous and is actually just in the starting blocks.
I just summarized my own experiences on this in a blog article at https://peterneufeld.wordpress.com/2023 ... evitation/.
Have a nice weekend.

[Fernando Perez]

Peter, your blog on this subject is very revealing.
But I am still afraid that, for mere mental comfort, we will stop exercising our reasoning in programming and, little by little, we will be incapable of solving problems by ourselves.
As a curiosity: If in https://www.perplexity.ai/ I ask: Program in annex RDS to control a led, I get an answer (and a valid one).
If I do the same in ChatGPT (chat.openai.com), it tells me that it doesn't know what Annex is and goes for Arduino.

I'm sorry Francesco.

[PeterN]

Fernando, I have to agree again

But times are changing, and I must admit that today we have reached a state where people without any knowledge of internal functions are very well supported by IT systems in their work. Hopefully, they will not lose this work now in the very next phase.

I may also report a (pleasant) conversation with my radiologist. He was quite optimistic that his 30 years of professional experience in differentiating cancer-suspicious from non-suspicious scans could soon be well supported by an AI. This would allow even inexperienced people to save lives. However, it is to be feared that these "small" radiologists might not be able to gain as much professional experience as he did. Will this knowledge then disappear one day, or only be available in silicon!

This will remain a very complex issue!!!

[cicciocb]

Peter, your blog on this subject is very revealing.
But I am still afraid that, for mere mental comfort, we will stop exercising our reasoning in programming and, little by little, we will be incapable of solving problems by ourselves.
As a curiosity: If in https://www.perplexity.ai/ I ask: Program in annex RDS to control a led, I get an answer (and a valid one).
If I do the same in ChatGPT (chat.openai.com), it tells me that it doesn't know what Annex is and goes for Arduino.

I'm sorry Francesco.

I know, I already tried some time ago ...
But I gave to chatgpt the URL of the help file and he was able to give me some tricks

[PeterN]

Peter, your blog on this subject is very revealing.
...
As a curiosity: If in https://www.perplexity.ai/ I ask: Program in annex RDS to control a led, I get an answer (and a valid one).
If I do the same in ChatGPT (chat.openai.com), it tells me that it doesn't know what Annex is and goes for Arduino.

Thank you for this food for thought. Was very interesting to look at the results of the same questions after this self-disclosure:

ChatGPT and Perplexity are both AI tools, but they differ in functionality due to a difference in design philosophy. Perplexity AI is a search engine that uses artificial intelligence to find and summarize online data. On the other hand, ChatGPT is an AI text generator that converses almost like a person and writes human-like text
...