Test on Open AI Just for Fun

General FreeBASIC programming questions.
Post Reply
exagonx
Posts: 315
Joined: Mar 20, 2009 17:03
Location: Italy
Contact:

Test on Open AI Just for Fun

Post by exagonx »

Hello friends.
This morning, at the suggestion of a colleague of mine, I wanted to analyze the code written in FreeBASIC with OpenAI to see if it would suggest a better method.

Well I admit that it generated some remarkably elegant and very neatly written code, but it doesn't work at all, and this is strange given that I inserted a function compatible with VisualBASIC and VB.NET. Apparently OpenAI, as much as artificial intelligence is not able to understand abstract algorithms used for data encoding.

Mankind vs AI 1 - 0
caseih
Posts: 2158
Joined: Feb 26, 2007 5:32

Re: Test on Open AI Just for Fun

Post by caseih »

Yeah I asked OpenAI for some example code in FreeBASIC to use GTK3, and it was way off. Kept using an undefined constant "nullptr" or something. I pointed out that nullptr was not defined and it said, oh you're right, I'm sorry I'll try again, and spit out the exact same code again. After doing this a couple of times it literally went off in a huff and said it didn't want to talk to me anymore. Hilarious. Of course it can only come up with stuff as good as humans have come up with in the past, and FreeBASIC example code on the net is in short supply. AI certainly does not know how to program, nor can it learn new languages the way humans do from looking at a language spec. OpenAI is remarkable at faking knowledge, wisdom, and intelligence. Not sure what this says about our own wisdom and intelligence.
exagonx
Posts: 315
Joined: Mar 20, 2009 17:03
Location: Italy
Contact:

Re: Test on Open AI Just for Fun

Post by exagonx »

caseih wrote: Sep 10, 2023 16:28 Not sure what this says about our own wisdom and intelligence.
I remember that a few years ago... in 1997 a synaptic architecture processing system with more than 1000 PCs was completed, the system was remarkably powerful because using many PCs connected to the network it was able to carry out considerably complex processing by optimizing the packets and creating routes optimal with the fastest connections, furthermore the work was divided based on the number of connected elements, in 1998 an attempt was made to start a self-learning procedure, in 1999 the whole project was dismantled and part of the code was handed over to SETI which has developed an application that processes data received from space.
Some argue that it was too close to real artificial intelligence.
Unfortunately they are only hypotheses, but I like the idea of being able to develop a code capable of learning, I personally am too ignorant to go that far, I still think with "If then" and in more complex cases I add "Else"
Post Reply