Chat GPT

[oddemann]

I got Chat GPT to make me a Basic game!

10 RANDOMIZE

20 NUMBER = INT(RND(1)*10) + 1

30 PRINT "I'm thinking of a number between 1 and 10."

40 INPUT "What is your guess? ", GUESS

50 IF GUESS = NUMBER THEN GOTO 90

60 PRINT "Sorry, that's not the number I was thinking of."

70 PRINT "Try again!"

80 GOTO 40

90 PRINT "Congratulations! You guessed the number!"

100 END

[oddemann]

With spaceship...

 

Spoiler

10 REM spaceship game
20 PRINT "Welcome to the spaceship game!"
30 PRINT "You are the pilot of a spaceship."
40 PRINT "You have to navigate through a asteroid field and reach the end of the game."
50 PRINT "Use the arrow keys to move your spaceship left, right, up, or down."
60 PRINT "Be careful, if you hit an asteroid, you lose."
70 PRINT "Good luck!"
80 GOTO 100

100 REM initialize variables
110 POSITIONX = 20
120 POSITIONY = 20
130 ASTEROIDX = INT(RND(1)*40) + 1
140 ASTEROIDY = INT(RND(1)*40) + 1
150 SCORE = 0

200 REM draw spaceship and asteroid
210 GRAPHICS 1
220 SETCOLOR 2,1,1
230 PLOT POSITIONX, POSITIONY
240 SETCOLOR 3,1,1
250 PLOT ASTEROIDX, ASTEROIDY
260 SETCOLOR 0,0,0
270 DRAWTO POSITIONX+1, POSITIONY
280 DRAWTO POSITIONX, POSITIONY+1
290 DRAWTO POSITIONX-1, POSITIONY
300 DRAWTO POSITIONX, POSITIONY-1
310 DRAWTO POSITIONX+1, POSITIONY

400 REM get user input
410 GETKEY k
420 IF k = "up" THEN POSITIONY = POSITIONY-1
430 IF k = "down" THEN POSITIONY = POSITIONY+1
440 IF k = "left" THEN POSITIONX = POSITIONX-1
450 IF k = "right" THEN POSITIONX = POSITIONX+1

500 REM check for collision
510 IF POSITIONX = ASTEROIDX AND POSITIONY = ASTEROIDY THEN GOTO 900

600 REM update score
610 SCORE = SCORE + 1
620 PRINT "Score: "; SCORE
630 GOTO 200

900 REM game over
910 PRINT "Game over! You hit an asteroid."
920 PRINT "Your final score is: "; SCORE
930 END

 

 

[oddemann]

It dos not work, but impressive! Poking it more... could it make a cool game? Impressive even if it dos not work, guess some clean-up and it will work.

 

[pixelpedant]

Now the question is, can it be convinced to write a TI BASIC game.  Understanding that TI BASIC and TI-BASIC are unrelated dialects, and so it must attend to this distinction. 

[oddemann]

8 minutes ago, pixelpedant said:

Now the question is, can it be convinced to write a TI BASIC game.  Understanding that TI BASIC and TI-BASIC are unrelated dialects, and so it must attend to this distinction. 

This is a start... And I guess that the more we poke it, the more it will learn and make perfect over time. Even if there is TI BASIC and TI-BASIC, it is interesting that we are at the point where we can almost have a computer make a computer program on request!

[jrhodes]

I don't need A.I. to make me lazier... I have alcohol and the burden of life in general to do that for me

[oddemann]

3 hours ago, jrhodes said:

I don't need A.I. to make me lazier... I have alcohol and the burden of life in general to do that for me

lol, I see that point... For me... the point of a computer IS the make it easier. To find ways to use the computer to do some stuff automatically. This is just a new level of how to use the computer.
AND... at the same time... I see how this could kill AND give time for more creativity.
Also the hole A.I. question is now a real thing.

The time of "I'll be back" is getting closer, or? lol

[+OLD CS1]

35 minutes ago, oddemann said:

The time of "I'll be back" is getting closer, or? lol

The AI industry is dead so long as the AI filter industry flourishes.

[oddemann]

Well I cant but test it and have fun...

So we have Lotto here in Norway... 7 numbers... Evil plan... I will become RICH hehehehe (lol in a evil crazy way) and I will not share it with ChatGPT!

YES it dos not know all the nuances, BUT I can fix that. I am just interested if I can make it do the iterations I ask of it. And I think it dos great!

"Make it a program generating 7 numbers from 1 to 34"
 

Spoiler

10 RANDOMIZE
20 FOR I = 1 TO 7
30   NUMBER = INT(RND(1)*34) + 1
40   PRINT NUMBER
50 NEXT I
60 END

 

OK, Nice! And easy!

But my evil mind keept going...

"Make it generate 7 numbers a 1000000 times and then avg 7 numbers"
 

Spoiler

10 RANDOMIZE
20 FOR I = 1 TO 1000000
30   FOR J = 1 TO 7
40     NUMBER = INT(RND(1)*34) + 1
50     SUM = SUM + NUMBER
60   NEXT J
70   AVERAGE = SUM / 7
80 NEXT I
90 PRINT "The average of the generated numbers is: "; AVERAGE
100 END

 

OK, did what I asked for... Bad question bad program... BUT... Surprise surprise "ChatGPT: You can also store those numbers in an array and then use the array to calculate the average." PS! It also told me some miner stupid info about something taking time...

My evil mind is not all there, but me and ChatGPT, together... can't be stopped!
 

So I told ChatGPT, "store those numbers in an array and then use the array to calculate the average"
 

Spoiler

10 RANDOMIZE
20 DIM NUMBERS(7)
30 FOR I = 1 TO 1000000
40   FOR J = 1 TO 7
50     NUMBER = INT(RND(1)*34) + 1
60     NUMBERS(J) = NUMBER
70   NEXT J
80   FOR K = 1 TO 7
90     SUM = SUM + NUMBERS(K)
100   NEXT K
110 NEXT I
120 AVERAGE = SUM / (7 * 1000000)
130 PRINT "The average of the generated numbers is: "; AVERAGE
140 END

 

 

I have not tested it, for sure buggy! But the idea that I had, I could reason out with ChatGPT and come out with a pretty good starting point. If this would work on the TI... probably not. But... If you can do this with TI Basic... then soon it will be open season for people with great imagination to start to make programmes never seen before, much easier!

This is the start, much as the computer it self started out! We are playing with a computer that is JUST 40 years old... what will happen with ChatGPT in 40 years. (I guess it will take less time to be "perfect".) I love the beginning. Even had it make me a hypnotic script... Not the best, but total useful. Even got it to make me a DnD story, and then with a twist. And for a "I need something new NOW!", great! Loved the name it gave the Dragon... "Smoky the Dragon!" hahahahahaha I will for SURE use that!

I see limitations NOW... But I also see the potential!

 

Edited January 28, 2023 by oddemann

[+TheBF]

6 hours ago, oddemann said:

Well I cant but test it and have fun...

So we have Lotto here in Norway... 7 numbers... Evil plan... I will become RICH hehehehe (lol in a evil crazy way) and I will not share it with ChatGPT!

YES it dos not know all the nuances, BUT I can fix that. I am just interested if I can make it do the iterations I ask of it. And I think it dos great!

"Make it a program generating 7 numbers from 1 to 34"
 

  Hide contents

10 RANDOMIZE
20 FOR I = 1 TO 7
30   NUMBER = INT(RND(1)*34) + 1
40   PRINT NUMBER
50 NEXT I
60 END

 

OK, Nice! And easy!

But my evil mind keept going...

"Make it generate 7 numbers a 1000000 times and then avg 7 numbers"
 

  Hide contents

10 RANDOMIZE
20 FOR I = 1 TO 1000000
30   FOR J = 1 TO 7
40     NUMBER = INT(RND(1)*34) + 1
50     SUM = SUM + NUMBER
60   NEXT J
70   AVERAGE = SUM / 7
80 NEXT I
90 PRINT "The average of the generated numbers is: "; AVERAGE
100 END

 

OK, did what I asked for... Bad question bad program... BUT... Surprise surprise "ChatGPT: You can also store those numbers in an array and then use the array to calculate the average." PS! It also told me some miner stupid info about something taking time...

My evil mind is not all there, but me and ChatGPT, together... can't be stopped!
 

So I told ChatGPT, "store those numbers in an array and then use the array to calculate the average"
 

  Hide contents

10 RANDOMIZE
20 DIM NUMBERS(7)
30 FOR I = 1 TO 1000000
40   FOR J = 1 TO 7
50     NUMBER = INT(RND(1)*34) + 1
60     NUMBERS(J) = NUMBER
70   NEXT J
80   FOR K = 1 TO 7
90     SUM = SUM + NUMBERS(K)
100   NEXT K
110 NEXT I
120 AVERAGE = SUM / (7 * 1000000)
130 PRINT "The average of the generated numbers is: "; AVERAGE
140 END

 

 

I have not tested it, for sure buggy! But the idea that I had, I could reason out with ChatGPT and come out with a pretty good starting point. If this would work on the TI... probably not. But... If you can do this with TI Basic... then soon it will be open season for people with great imagination to start to make programmes never seen before, much easier!

This is the start, much as the computer it self started out! We are playing with a computer that is JUST 40 years old... what will happen with ChatGPT in 40 years. (I guess it will take less time to be "perfect".) I love the beginning. Even had it make me a hypnotic script... Not the best, but total useful. Even got it to make me a DnD story, and then with a twist. And for a "I need something new NOW!", great! Loved the name it gave the Dragon... "Smoky the Dragon!" hahahahahaha I will for SURE use that!

I see limitations NOW... But I also see the potential!

 

You might want to tell it to never use a variable called NUMBER or NUM.

NUMBER is a reserved word in TI BASIC.

If you change NUMBER to N and change RND(1)   to RND it runs... for a very very long time.

 

 

[+mizapf]

Could be interesting to see whether it can create a program that outputs a random permutation of a given list of numbers. This may sound simple at first, but when you try it you will see that this is not like throwing dice. If it should have a chance, it must find some algorithm on the net.

[GDMike]

On 1/28/2023 at 10:49 AM, jrhodes said:

I don't need A.I. to make me lazier... I have alcohol and the burden of life in general to do that for me

Just tell it that you don't need no stupid AI and it'll probably tell you do it your damn self it's that smart- of an ass. Lol

[unhuman]

I started a similar thread for TI based development...  It did not go well:

 

[oddemann]

On 1/29/2023 at 5:28 AM, TheBF said:

You might want to tell it to never use a variable called NUMBER or NUM.

NUMBER is a reserved word in TI BASIC.

If you change NUMBER to N and change RND(1)   to RND it runs... for a very very long time.

 

 

I understand that this is not perfect, that is not what I am talking about.

What I see is a possibility to as an example, take all the info about the TI, feed it into the ChatGPT and it will slowly be the one you can ask anything. Imagine this "engine" read all the info in this forum and then all the books, everything online. Then it would be a great teacher, you would feed it your program and it would debug it for you.

So I am viewing this from the possibility, not how perfect it is now. Some changes and let it learn, it could be a super cool tool. I see great possibilities. Also I see how ChatGPT could answer beginner question WITH endless patients for slow or just new learners.

So, the details, I can fix! But I got a great structure of a program in seconds! Also this is JUST the beginning.

[oddemann]

17 hours ago, unhuman said:

I started a similar thread for TI based development...  It did not go well:

 

I am a little surprised of the "ChatGPT is not cool" attitude. I see this as a new step in the computer evolution. A program that we make is a really stupid "A.I." to play a game. For me this is the natural evolution of computers, where we get it to, again do stuff that takes time to do. More can be automated. I think that this tool might be for the people that can imagine ways of making stuff in a new way. The danger I think is that this might make us even more engrossed in the virtual world.

One could argue that this makes us stupid, well... that is what the computer dos all over! I can use a computer to write a document and make it look nice! I could not do that by hand, unless I use a lot of time to be good at it.

So I am a little surprised that there is so little interest in ChatGPT. BUT... usually we are "scared" of the new stuff, so in that sense it is understandable.

In my mind, I need to restructure my mind so see how I can use this tool to become even better! It makes me tingling with, "Wow, what more can I do with this?"

This can't be stopped, the "plague" is out of the box. Now we need to grow to tame it! I am more scared of mindless use of this possibility! (T-1000 vs Humans - by Skynet)

[CapitanClassic]

1 hour ago, oddemann said:

So I am a little surprised that there is so little interest in ChatGPT. BUT... usually we are "scared" of the new stuff, so in that sense it is understandable.

I don’t think people are scared of ChatGPT, and most of the fear-mongers saying this is the end of white-collar jobs are vastly overreacting.  I have less interest in ChatGPT than the AI programs that generate new unique artwork. The artwork that it generates look interesting, and I could see it being used by good artists to increase their level of productivity. (Although the raw output is never perfect, see hands which the AI is particularly bad with)

 

The ChatGPT program is mostly just a search engine with a decent handling of natural language reading/writing. This is just an evolution of the old Eliza the Computer Psychologist program. Just ask it questions in an area of knowledge that you are an expert in, and you will see how wrong the AI can be (Stack Exchange has banned ChatGPT answers due to their lack of quality.) Not only that, but similar prompts give responses that seem sometimes canned, it also doesn’t seem to grasp humor.

 

What ChatGPT and similar programs will do is exacerbate the productivity differences that already exist between people (Pareto principle), makes smart industrious people even more productive (you will still need people, just fewer of them). There will be AI programs that specialize in contract law that make lawyers more productive, and AI programs that specialize in writing software that will make boilerplate plate code that programmers can specialize and fix. This will result in a better world in the long run, with fewer people doing meaningless work.

[+TheBF]

3 hours ago, oddemann said:

I understand that this is not perfect, that is not what I am talking about.

What I see is a possibility to as an example, take all the info about the TI, feed it into the ChatGPT and it will slowly be the one you can ask anything. Imagine this "engine" read all the info in this forum and then all the books, everything online. Then it would be a great teacher, you would feed it your program and it would debug it for you.

So I am viewing this from the possibility, not how perfect it is now. Some changes and let it learn, it could be a super cool tool. I see great possibilities. Also I see how ChatGPT could answer beginner question WITH endless patients for slow or just new learners.

So, the details, I can fix! But I got a great structure of a program in seconds! Also this is JUST the beginning.

I agree. These things are already changing the world. 

Unfortunately we can't train ChatGPT as it exists now . It is locked at 2021 when it stopped training.

The day will come when these things are open and learning all the time. That will usher in a new era ... for better or worse.  

[unhuman]

On 1/31/2023 at 1:51 PM, oddemann said:

I am a little surprised of the "ChatGPT is not cool" attitude. I see this as a new step in the computer evolution. A program that we make is a really stupid "A.I." to play a game. For me this is the natural evolution of computers, where we get it to, again do stuff that takes time to do. More can be automated. I think that this tool might be for the people that can imagine ways of making stuff in a new way. The danger I think is that this might make us even more engrossed in the virtual world.

One could argue that this makes us stupid, well... that is what the computer dos all over! I can use a computer to write a document and make it look nice! I could not do that by hand, unless I use a lot of time to be good at it.

So I am a little surprised that there is so little interest in ChatGPT. BUT... usually we are "scared" of the new stuff, so in that sense it is understandable.

In my mind, I need to restructure my mind so see how I can use this tool to become even better! It makes me tingling with, "Wow, what more can I do with this?"

This can't be stopped, the "plague" is out of the box. Now we need to grow to tame it! I am more scared of mindless use of this possibility! (T-1000 vs Humans - by Skynet)

Oh, that wasn't my approach at all.  ChatGPT was fun to play with.  My frustration was how it is not relevant for TI usage, yet.  Hopefully, they actually teach it so that it could be useful in this case.  It's been useful in other, real world cases, for me.