Search the Community
Showing results for tags 'ai'.
-
The current mania over Artificial Intelligence (AI) reminded me of a program I wrote for the original Atari 800 in 1984, for a post-graduate private computer school class. I've left the manual describing the program completely as is, except for redacting my old address on the cover page. Apologies for the yellowing pages, which are 39 years old now. The slight skewing from the scanner didn't lose any characters in the coding section and you can still type in the full code if you desire. I submit this as open-source and as a learning tool, to see how a basic - also BASIC - AI program was written in the very early days, almost before the IBM PC came out. It is based upon a very simple AI program in Compute! magazine - just 23 lines, or <1/10th the size of my greatly enhanced program. The program was originally designed for the Atari 800 but ought to run on the Atari 1450xld with speech options. The 1450xld was never officially released before Atari declared bankruptcy, but I was able to find the BASIC code that would have worked for that machine, and included it in this late stage Atari program, completed by May, 1984. Superlearner also has an option to print what it has learned to a printer, or save the output to a disk or cassette. The Atari had no access to the internet, which, in any case, did not exist in 1984, so there was no option for a BASIC AI program except for the user to manually enter facts and non-facts about subjects for the AI program to mull over. Menu options make this a simple matter. Everything is menu-driven, including the sample data already built into the program. I lost the function of my Atari 410 cassette recorder a long time ago and although I have a working Atari 810 disk drive, I have not had the patience to type in the program again. If there is sufficient interest in the program I may do so again, and provide some sample output from the program. There are some similarities to today's AI but of course, the level of complexity and capacity is millions of orders of magnitude greater today. I present this as a simple example of AI and of historical background to today's AI apps. Enjoy! Superlearner AI Manual and Program_Redacted.pdf
- 11 replies
-
- 9
-
- ai
- artificial intelligence
-
(and 2 more)
Tagged with:
-
Hi there, I decided to rescue this program using scan and OCR-ed text file but I'm not sure if everything is OK there. All details in this topic: Yeah, linking it here because it's Atari 80bit related after all.
-
(X-posted from Atari General) The current mania over Artificial Intelligence (AI) reminded me of a program I wrote for the original Atari 800 in 1984, for a post-graduate private computer school class. I've left the manual describing the program completely as is, except for redacting my old address on the cover page. Apologies for the yellowing pages, which are 39 years old now. The slight skewing from the scanner didn't lose any characters in the coding section and you can still type in the full code if you desire. I submit this as open-source and as a learning tool, to see how a basic - also BASIC - AI program was written in the very early days, almost before the IBM PC came out. It is based upon a very simple AI program in Compute! magazine - just 23 lines, or <1/10th the size of my greatly enhanced program. The program was originally designed for the Atari 800 but ought to run on the Atari 1450xld with speech options. The 1450xld was never officially released before Atari declared bankruptcy, but I was able to find the BASIC code that would have worked for that machine, and included it in this late stage Atari program, completed by May, 1984. Superlearner also has an option to print what it has learned to a printer, or save the output to a disk or cassette. The Atari had no access to the internet, which, in any case, did not exist in 1984, so there was no option for a BASIC AI program except for the user to manually enter facts and non-facts about subjects for the AI program to mull over. Menu options make this a simple matter. Everything is menu-driven, including the sample data already built into the program. I lost the function of my Atari 410 cassette recorder a long time ago and although I have a working Atari 810 disk drive, I have not had the patience to type in the program again. If there is sufficient interest in the program I may do so again, and provide some sample output from the program. There are some similarities to today's AI but of course, the level of complexity and capacity is millions of orders of magnitude greater today. I present this as a simple example of AI and of historical background to today's AI apps. Enjoy! Superlearner AI Manual and Program_Redacted.pdf
-
- 8
-
- ai
- artificial intelligence
-
(and 2 more)
Tagged with:
-
In my code, I have an internal timer that when it runs out will gosub to what is supposed to be something that will change what direction player 2 is pointing and then reset the timer. The problem with my code is that the rand will only work once. The timer setup: dim Timer1=f.g dim Timer2=h.j Timer1=20 Timer2=1 main Timer1=Timer1-Timer2 if Timer1=2 then gosub randomgenerator:Timer1=20 The random selector: randomgenerator if rand<20 && rand>0 then p1_orientation=1 if rand<40 && rand>19 then p1_orientation=2 if rand<60 && rand>39 then p1_orientation=3 if rand<80 && rand>59 then p1_orientation=4 if rand<100 && rand>79 then p1_orientation=5 if rand<120 && rand>99 then p1_orientation=6 if rand<140 && rand>119 then p1_orientation=7 if rand<160 && rand>139 then p1_orientation=8 if rand<180 && rand>159 then p1_orientation=9 if rand<200 && rand>179 then p1_orientation=10 if rand<220 && rand>199 then p1_orientation=11 if rand<240 && rand>219 then p1_orientation=12 return Thanks!
-
A box recently arrived at my house. In it was a handful of 2600 homebrews, including Medieval Mayhem, Gunfight, and Warring Worms. Now beforehand I considered myself a "pro" at games like PONG and Warlords. I was good enough at PONG aka "Video Olympics" that I was about an even match for my buddy using the difficulty switches to handicap myself, with me using the small paddle and him using the larger one. And I could whup him in Warlords, too. I remembered the AI being really easy to beat on the original Warlords, but when I popped in the Medieval Mayhem cart, I got my @ss handed to me on a silver platter. At first, I played a round with the default "fast" setting on 3 balls, then "slow", then "Medieval," getting beat every time, and after reading the manual, I set it to "Kids" with 1 ball, disabled holding on all but my own paddle, and set myself up as the Blue Castle, which seemed to be the strongest CPU player. I still kept getting whupped, and I couldn't win a single round, ever... My score (Medieval Mayhem, "kids", 1 ball): 0-1-4-5. It seemed like the "speed" setting only slows down the rate of the balls, but not the reaction of the AI players. Instead, the CPU players hit the ball every time with ninja reflexes, and the rounds drag on forever, sometimes with me (blue) eliminated before the red and green even had a single hit on their castles. I thought to myself, "I know I don't suck this bad," and their doesn't seem to be any option to actually "dumb down" the CPU opponents. It's like I'm playing the same "expert" opponents regardless of what speed I set the balls at. "kids" mode with 1 ball just made the gameplay painfully slow. I didn't recall sucking that bad, so I popped in the original Warlords and started a 1 player game. Like I expected, it was a total shutout within a few very short minutes. My score (Warlords, single player, variation #4): 5-0-0-0 While I must admit, the graphics and audio are absolutely superb, it's just not any fun unless I can get some kind of balanced difficulty. Next time my buddy comes over to my house, I'll give it a whirl, but it seems like there's no real way to dumb down the AI, short of playing a 2 player game in 4 player mode against "dead" opponents, which is boring and lame. I don't even own a second set of paddles. My friend is slightly less skilled than I am at Warlords (although he could whup my butt in Combat), and I imaging when we finally play each other on Medieval Mayhem, the CPU will probably annihilate us, unless we play doubles mode. Honestly, the dumb AI in the original Warlords made for a great multiplayer experience. In opposite corners, we would often play co-op until the enemies were annihilated, then it would be all out war between the two of us, winner take all. I don't know when I'll be able to invite him over again with his work schedule, and my mom and fiance are complete non-gamers. Anyone else run into problems trying to play Medieval Mayhem, or do I just really suck that bad? If I could somehow get the CPU opponents to be somewhere harder than Warlords, but easier than it is now, then maybe it would be a more well-rounded experience. I really want to be able to set the CPU so that I win some and lose some, but as it stands, Warlords is a pushover and Medieval Mayhem just seems unfair. Don't get me wrong, I want to enjoy this game, but it's hard to play against a CPU that constantly wins or constantly loses. I read the reviews and they said the AI was a lot smarter in Medieval Mayhem, but this is insane. I know Atari fixed the CPU difficulty in single player variations of PONG by setting a maximum velocity for the paddle to travel, ensuring that the human opponent would have a fair chance to win. I really want to love this game. Hopefully, the AI in "Gunfight" and "Warring Worms" will be a little more balanced...
-
This isn't specifically Intellivision related, but I thought I might mention it for the interested: Coursera has a number of free, online courses, covering everything from algorithms to calculus. I just signed up for an interesting-looking course on Game Theory. I've previously taken the free AI and Model Thinking courses among others. So far, I've enjoyed them all. If you're thinking about designing a simulation game world, such as you might find in Utopia, or want to program "smart" enemies in Chess-like games, or something else that involves decisions and modeling, these courses provide a lot of insights on how to attack the challenge. I admit, part of my inspiration to reverse-engineer Utopia came from learning about all the different ways one could model an economy and population in Model Thinking. I wanted to see how Don Daglow did it in his world to compare. :-) The Model Thinking course comes back around in September, it appears. The professor that taught the AI course I took started his own online venture, Udacity. I haven't taken any of the Udacity-branded courses yet. They haven't had anything in their catalog that piqued my interest. Anyway... enough "advertising", such that it is. (Really, I'm just enjoying all these great courses!)