|
|
|
|
| Welcome, Guest | Home | Search | Login | Register | |
| Author | Asking ChatGPT to write a program for System 7. (Read 7791 times) | ||||||||||||||
|
ReleaseTheGeese
32 MB ![]() ![]() ![]() Posts: 37 Division 1 Computer Support
|
on: December 18, 2022, 18:47
Having some fun with our future AI overlord. I asked ChatGPT, could you write me a program in Pascal for Macintosh System 7 that makes a simple window appear? I admit I was trying to stump it. Surprisingly it spat out the following: Code: [Select] program SimpleWindow;Then it gave up at the end. Not bad, though I wonder how well it did? If any S7 pascalites care to chip in. I am more of a C-nile myself
|
||||||||||||||
|
68040
|
512 MB ![]() ![]() ![]() ![]() ![]() Posts: 950 68k - thy kingdom come, thy will be done !
Reply #1 on: December 18, 2022, 18:57
|
I was already stumped at ChatGPT AI(??)
|
Bolkonskij
|
Administrator 1024 MB ![]() ![]() ![]() ![]() ![]() Posts: 2023
Reply #2 on: December 19, 2022, 19:19
|
No real Pascal knowledge here. On a first glance, I can't seem to see it call the Toolbox routine CreateNewWindow() for opening a (b&w) window which is the essentially starting point for a Macintosh application. For whatever reason, it uses GetNewDialog() ? Maybe I don't get it, it's a possibility :-) I also don't see it properly initializing the Toolbox before that. Interestingly, it gets the 128 as resource ID right. Still, the ChatGPT AI hype gets a bit annoying lately. To me, in the end it's just another database with a lot of nested IFs to find the info you're looking for. Feed it, it will thrive. Starve it, it will fail. Much less a revolution, than just an evolution.
Last Edit: December 19, 2022, 19:57 by Bolkonskij
|
68040
|
512 MB ![]() ![]() ![]() ![]() ![]() Posts: 950 68k - thy kingdom come, thy will be done !
Reply #3 on: December 19, 2022, 20:56
|
A typical chat bot "AI" answers complex questions with some gibberish it snatched from the net or previous user input. Those AI engines outsource the "sense making" part to the brain of the observer, by guessing what (s)he is looking for and presenting any text snippet that might fit that assumption. Its like a Psychic on steroids, relying on the human to assume intelligence, where there are nothing more than blind algorithms at work, performing statistical guesswork and throwing text sequences back at you whose content they don't understand. Most readers will not be able/willing to do the detailed analysis Bolkonskij has done and instead just conclude that there must be a real intellect behind it. And according to the Eliza rules, as soon as an AI engine can fool the observer into assuming its intelligent, the engine is assumed to be intelligent. So we hand more and responsibilities over to code sequences, that are doing sophisticated statistical correlations - and nothing more. Yet the problem with statistical hits is that they are mostly on target - mostly.
Last Edit: December 19, 2022, 20:58 by 68040
|
|
Pages: [1]
|
| ||||
|
© 2021 System7Today.com. |



