ChatGPT
#11
Before long they would need more servers, although not as rapid as the growth of "archive-dot-org" or Wikipedia. Hooray binary search and heuristics! The authors must be working hard on something that could pick up good results from searches on Google, to solve a problem like addition or update of keywords/functions to QB64PE.
Reply
#12
Quote:ChatLGBT is nothing but fancy autocomplete. There, you heard my side first, ha.


Goood one STATIC Big Grin
Reply
#13
So easy to dismiss with an oversimplified excuse. Hey if your plate is full, good for you!

I went a couple steps further in signing up for ChatGPT but quit when asked for phone number.

Why do you suppose they would need that?

I see PLUS cost 20$ a month.
b = b + ...
Reply
#14
(02-23-2023, 08:19 PM)bplus Wrote: I went a couple steps further in signing up for ChatGPT but quit when asked for phone number.

Why do you suppose they would need that?
To sell, of course. Telemarketing.
Ask me about Windows API and maybe some Linux stuff
Reply
#15
They want the phone number to sell your data to marketers so they make more money to keep the servers running.
Reply
#16
(02-23-2023, 08:19 PM)bplus Wrote: So easy to dismiss with an oversimplified excuse. Hey if your plate is full, good for you!

I went a couple steps further in signing up for ChatGPT but quit when asked for phone number.

Why do you suppose they would need that?

I see PLUS cost 20$ a month.

They didn't ask for my phone number? I signed up through OpenAI though.
Software and cathedrals are much the same — first we build them, then we pray.
QB64 Tutorial
Reply
#17
(02-22-2023, 05:26 AM)mnrvovrfc Wrote: Despite what I said earlier, I was a bit amused with how a couple of other members of this forum interacted with it. Like the first thread by madscijr that I read about it, to build some text-only adventure LOL.

First I tried to get it to implement an associative array, and it worked. 
Then I asked it to optimize it with hashing, but it only showed part of the solution. 
It seems it only outputs a page at a time! 
I had to ask it to show the rest of the program, and it did!

The one-page-at-a-time problem got ridiculus when I asked it to make a 2-player Spacewar! game
I had to ask it "show the rest" a total of 21 times!!! 
And each time, it was as if it had re-created the entire solution from scratch, 
and it was slightly different than the previous time. 
Seems if you ask it to do something multiple times, it might not do it exactly the same each time. 
So piecing together all the code that it spat out was a little disjointed. 
Yet another unfinished project was fixing it to get it to actually run (yeah right). 

Does anyone know if there is some way to get it to output the whole answer at once, or do you have to sign up for a paid account? 


has anyone had ChatGPT write QB64 code yet?
wrangling chatgpt to get it to write a more complex program - Spacewar!
Reply
#18
(02-23-2023, 08:56 PM)TerryRitchie Wrote:
(02-23-2023, 08:19 PM)bplus Wrote: I went a couple steps further in signing up for ChatGPT but quit when asked for phone number.
Why do you suppose they would need that?
I see PLUS cost 20$ a month.
They didn't ask for my phone number? I signed up through OpenAI though.

Same here, I signed up with OpenAI a while back (before ChatGPT). 
I wonder if that PLUS removes the one-page answer limitation, or provides some privacy protection? 
(Not counting on it, but it's worth asking!)
Reply
#19
(02-23-2023, 10:39 PM)madscijr Wrote:
(02-23-2023, 08:56 PM)TerryRitchie Wrote:
(02-23-2023, 08:19 PM)bplus Wrote: I went a couple steps further in signing up for ChatGPT but quit when asked for phone number.
Why do you suppose they would need that?
I see PLUS cost 20$ a month.
They didn't ask for my phone number? I signed up through OpenAI though.

Same here, I signed up with OpenAI a while back (before ChatGPT). 
I wonder if that PLUS removes the one-page answer limitation, or provides some privacy protection? 
(Not counting on it, but it's worth asking!)

I had to do the same thing, ask it to show me multiple pages. I have no idea if the plus versions get rid of this limitation (or bug).

It's fun to play around with none-the-less. I asked it for a new invention idea. It told me that solar powered drones would be a good bet. I asked it for a name for my new brown kitten. It said Mocha. I asked it to respond more like a human, and it did!

Nutty stuff. I have lots of little code snippets it made for me too. Interesting times we live in.
Software and cathedrals are much the same — first we build them, then we pray.
QB64 Tutorial
Reply
#20
(02-23-2023, 10:54 PM)TerryRitchie Wrote:
(02-23-2023, 10:39 PM)madscijr Wrote:
(02-23-2023, 08:56 PM)TerryRitchie Wrote: They didn't ask for my phone number? I signed up through OpenAI though.

Same here, I signed up with OpenAI a while back (before ChatGPT). 
I wonder if that PLUS removes the one-page answer limitation, or provides some privacy protection? 
(Not counting on it, but it's worth asking!)

I had to do the same thing, ask it to show me multiple pages. I have no idea if the plus versions get rid of this limitation (or bug).

It's fun to play around with none-the-less. I asked it for a new invention idea. It told me that solar powered drones would be a good bet. I asked it for a name for my new brown kitten. It said Mocha. I asked it to respond more like a human, and it did!

Nutty stuff. I have lots of little code snippets it made for me too. Interesting times we live in.

What we should really be doing is asking it to cure diseases, improve everyone's mental health & fix social problems, figure out how to slow down aging, come up with ways to prevent war and famine, clean up pollution, and improve on the economic & government systems, so no one goes hungry, dies too soon, kills each other, robs each other, or sponges off the system too much. 

Of course it won't come up with perfect answers - what it comes up with would have to be vetted by real humans, but it may contain some valuable insights we might not have arrived at so soon. 

Perhaps if we put a few thousand instances on the job, and have a bunch of them double-checking and verifying each others' answers, the results will be better. 

Like giving a million monkeys a typewriter, where eventually one will produce "Hamlet", except these monkeys can read and verify that the other monkey's "Hamlet" is real Shakespeare. 

Can they weed out bad solutions using the scientific method and arrive at the top 3 for a given problem after so many cycles? 

The down side of all this, of course, is this stuff can also be used to come up with more efficient weapons and tactics to do not-so-nice things. We can be sure that all kinds of corporations, governments, militaries, investors, hackers, mad scientists, and everyone else are throwing whatever they have at it, to try and get a leg up. So there's that. 

It's like that old Chinese curse, "May you live in interesting times"! We sure are!! LoL
Reply




Users browsing this thread: 11 Guest(s)