4 ms·

A GPT version of parser adventures would be awesome.

by rado 3y ago

palad1n 3y ago
Like AI Dungeon?
anthk 3y ago
It would suck a lot. GPT lacks coherence in lots of fronts. A game made with Inform6 (OOP) or Inform 7 (natural language), even done by an amateur it's much better on consistency and playability.
- ilaksh 3y ago
  You could use a similar engine to keep track of things and feed them to GPT could provide new descriptions or ideas, or use functions provided by the engine to modify map state or run appropriate commands according to the what the user says. To start, it could be as simple as giving GPT the list of valid commands for the user state and location and having it translate if the user asks for something that is equivalent but doesn't match the command exactly.
SketchySeaBeast 3y ago
Maybe. The problem is that it could hallucinate it's way into a corner and kill the intended narrative as it creates non-sensical objects and characters. But it is very funny to think of a system requiring massive amounts of compute being used to try and replicate a genre as old as PCs itself.
- DennisP 3y ago
  My girlfriend uses Siri to add up numbers. That's a massive amount of compute to replicate a mechanical adding machine.
- anthk 3y ago
  A GTX 3080 Golden Edition TM vs a Pentium MMX/32MB with Inform6 (and that's more than enough to run several builds in parallel) or a Pentium II@333 with Inform7 and some text editor. More than funnny, it's shameful. And a huge waste of resources. I mean, if they used AI's for parsing papers helping bioinformatics and medicine, that would be a great case. Not for text adventures, where with Inform6 and basic logic you can create a game in a month. A week to write a good script (literarian script), another one to code it and two weeks polishing things and bugs here and there.
  - DennisP 3y ago
    Your brain does far more computation and you're saying to use it for a good part a month to do something an AI can do on the fly in seconds.
    - anthk 3y ago
      Tell me you never played IF (even the basic ones) without telling me so explicitly. If you play Entangled, Anchorhead, All Things Devours... the gameplay it's far above an AI can do. In ...devours... you have a time travelling puzzle where you must avoid a paradox with yourself. Good luck with that. And that's a game that will run under Frotz from DOS/Windows 95/old Linux/NetBSD in a 486. Or your basement Raspberry Pi, without even sweating a cycle. Put a Pentium MMX or a basic RpiA, play Galatea https://www.ifwiki.org/Galatea https://www.ifwiki.org/Galatea or Counterfeit Monkey https://www.ifwiki.org/Counterfeit_Monkey https://www.ifwiki.org/Counterfeit_Monkey and then enforcing an AI to create a coherent game with narrativeness, feelings and drama (without being formulaic on speech or narrative) without shitting itself over a few iterations of gameplays it's a dauntisk task. And these are text games, I repeat, an Amiga, Atari or a 486 could run without too much hazzle. Yet they are created with trivial tools also runable in a Pentium machine from 1997. Yet an AI can't reach these levels of gameplay even if they tried. Because maintaing coherence on described media with full of twists and easy to create "hard" puzzles like nothing (for an Inform6 progrmmer) it's a very hard task for a machine having to do plays on semantics.
      - DennisP 3y ago
        Ok, you're right that I haven't played a text game since I had a DOS machine, and now I'm intrigued. On the other hand, I do find myself a bit skeptical that such a sophisticated game is likely to be written as quickly as you describe. If a game designed in a week is more basic, then perhaps that would be doable by a good LLM, either now or in the near future.
        anthk 3y ago
        Well, basically in the 90's fans of the ZMachine interpreter from Zork/Infocom created games with amateur tools and a really easy OOP lang called Inform, which the latest OOP iteration it's Inform 6 and they made jewels such as Curses!, Jigsaw, Anchorhead... surpassing the quality of the infocom games, even creating an eigth (v8) version of the Zmachine, surpassing v3 and v5 editions from Infocom. One of the games features even time travelling puzzles having to fight with yourself and your previous steps you did in the game. That could make an LLM parser explode. Yet, you have the Inform6 source code at https//jxself.org/git/devours and you can compile and run it with inform6 on systems as small as a 486 or a Pentium with the inform6 compiler. Later with Inform 7 they created an English like language, describing the game by just using English like sentences. Emily Short created Galatea with Inform6 which is more a conversation experiment than a game, think of Eliza but with stereoids, having even "feelings" and "topics" on the conversation. Depending on how you chat with a statue and what topics you uncover, after a long chat you will reach one ending or another. OFC the topic and the feelings, previous chat... it's dynamic and not pre-generated like a graphical adventure menu. Then there is Counterfeit Monkey, where you can turn an object into anagrams of others in order to solve clever puzzles, and the object count it's really big. That could be really hard to implement with an LLM.
spencerflem 3y ago
"There’s no question in my mind that such software could generate reasonably good murder mysteries, action thrillers, or gothic romances. After all, even the authors of such works will tell you that they are formulaic. If there’s a formula in there, a deep learning AI system will figure it out. Therein lies the fatal flaw: the output will be formulaic. Most important, the output won’t have any artistic content at all. You will NEVER see anything like literature coming out of deep learning AI. You’ll see plenty of potboilers pouring forth, but you can’t make art without an artist. This stuff will be hailed as the next great revolution in entertainment. We’ll see lots of prizes awarded, fulsome reviews, thick layers of praise heaped on, and nobody will see any need to work on the real thing. That will stop us dead in our tracks for a few decades. " from https://www.erasmatazz.com/library/interactive-storytelling/why-i-am-ending-further.html https://www.erasmatazz.com/library/interactive-storytelling/... (2018)

A GPT version of parser adventures would be awesome.

by rado 3y ago

A GPT version of parser adventures would be awesome.

palad1n 3y ago

Like AI Dungeon?

anthk 3y ago

It would suck a lot. GPT lacks coherence in lots of fronts. A game made with Inform6 (OOP) or Inform 7 (natural language), even done by an amateur it's much better on consistency and playability.

ilaksh 3y ago
You could use a similar engine to keep track of things and feed them to GPT could provide new descriptions or ideas, or use functions provided by the engine to modify map state or run appropriate commands according to the what the user says. To start, it could be as simple as giving GPT the list of valid commands for the user state and location and having it translate if the user asks for something that is equivalent but doesn't match the command exactly.

SketchySeaBeast 3y ago

Maybe. The problem is that it could hallucinate it's way into a corner and kill the intended narrative as it creates non-sensical objects and characters. But it is very funny to think of a system requiring massive amounts of compute being used to try and replicate a genre as old as PCs itself.

DennisP 3y ago
My girlfriend uses Siri to add up numbers. That's a massive amount of compute to replicate a mechanical adding machine.
anthk 3y ago
A GTX 3080 Golden Edition TM vs a Pentium MMX/32MB with Inform6 (and that's more than enough to run several builds in parallel) or a Pentium II@333 with Inform7 and some text editor. More than funnny, it's shameful. And a huge waste of resources. I mean, if they used AI's for parsing papers helping bioinformatics and medicine, that would be a great case. Not for text adventures, where with Inform6 and basic logic you can create a game in a month. A week to write a good script (literarian script), another one to code it and two weeks polishing things and bugs here and there.
- DennisP 3y ago
  Your brain does far more computation and you're saying to use it for a good part a month to do something an AI can do on the fly in seconds.
  - anthk 3y ago
    Tell me you never played IF (even the basic ones) without telling me so explicitly. If you play Entangled, Anchorhead, All Things Devours... the gameplay it's far above an AI can do. In ...devours... you have a time travelling puzzle where you must avoid a paradox with yourself. Good luck with that. And that's a game that will run under Frotz from DOS/Windows 95/old Linux/NetBSD in a 486. Or your basement Raspberry Pi, without even sweating a cycle. Put a Pentium MMX or a basic RpiA, play Galatea https://www.ifwiki.org/Galatea https://www.ifwiki.org/Galatea or Counterfeit Monkey https://www.ifwiki.org/Counterfeit_Monkey https://www.ifwiki.org/Counterfeit_Monkey and then enforcing an AI to create a coherent game with narrativeness, feelings and drama (without being formulaic on speech or narrative) without shitting itself over a few iterations of gameplays it's a dauntisk task. And these are text games, I repeat, an Amiga, Atari or a 486 could run without too much hazzle. Yet they are created with trivial tools also runable in a Pentium machine from 1997. Yet an AI can't reach these levels of gameplay even if they tried. Because maintaing coherence on described media with full of twists and easy to create "hard" puzzles like nothing (for an Inform6 progrmmer) it's a very hard task for a machine having to do plays on semantics.
    - DennisP 3y ago
      Ok, you're right that I haven't played a text game since I had a DOS machine, and now I'm intrigued. On the other hand, I do find myself a bit skeptical that such a sophisticated game is likely to be written as quickly as you describe. If a game designed in a week is more basic, then perhaps that would be doable by a good LLM, either now or in the near future.
      - anthk 3y ago
        Well, basically in the 90's fans of the ZMachine interpreter from Zork/Infocom created games with amateur tools and a really easy OOP lang called Inform, which the latest OOP iteration it's Inform 6 and they made jewels such as Curses!, Jigsaw, Anchorhead... surpassing the quality of the infocom games, even creating an eigth (v8) version of the Zmachine, surpassing v3 and v5 editions from Infocom. One of the games features even time travelling puzzles having to fight with yourself and your previous steps you did in the game. That could make an LLM parser explode. Yet, you have the Inform6 source code at https//jxself.org/git/devours and you can compile and run it with inform6 on systems as small as a 486 or a Pentium with the inform6 compiler. Later with Inform 7 they created an English like language, describing the game by just using English like sentences. Emily Short created Galatea with Inform6 which is more a conversation experiment than a game, think of Eliza but with stereoids, having even "feelings" and "topics" on the conversation. Depending on how you chat with a statue and what topics you uncover, after a long chat you will reach one ending or another. OFC the topic and the feelings, previous chat... it's dynamic and not pre-generated like a graphical adventure menu. Then there is Counterfeit Monkey, where you can turn an object into anagrams of others in order to solve clever puzzles, and the object count it's really big. That could be really hard to implement with an LLM.

spencerflem 3y ago

"There’s no question in my mind that such software could generate reasonably good murder mysteries, action thrillers, or gothic romances. After all, even the authors of such works will tell you that they are formulaic. If there’s a formula in there, a deep learning AI system will figure it out. Therein lies the fatal flaw: the output will be formulaic. Most important, the output won’t have any artistic content at all. You will NEVER see anything like literature coming out of deep learning AI. You’ll see plenty of potboilers pouring forth, but you can’t make art without an artist. This stuff will be hailed as the next great revolution in entertainment. We’ll see lots of prizes awarded, fulsome reviews, thick layers of praise heaped on, and nobody will see any need to work on the real thing. That will stop us dead in our tracks for a few decades. " from https://www.erasmatazz.com/library/interactive-storytelling/why-i-am-ending-further.html https://www.erasmatazz.com/library/interactive-storytelling/... (2018)