13 ms·

Emulating Pokemon Emerald on GPT-4

189 points by dangond 4y ago

frozenlettuce 4y ago
I played a RPG session with GPT-3 and it was quite fun. The problem was the combat part, as describing violent events is against the TOS (just when me and a party of villagers had prepared an attack against an orc encampment!) I guess that something that involves "animated violence" like pokemon might be a good idea to get around that.
- causality0 4y ago
  One interesting weak point I've noticed with GPT is that if you ask it to give examples from a category which in the real world has hundreds or thousands of different possibilities, it will give you the same half dozen answers every time. You ask it to give you "opponents" and it will always give you male humans. You ask it to give you "monsters" to kill and it always gives you orcs, giant spiders, skeletons. If a chest is supposed to have "useful items", it's always a healing potion.
  - asddubs 4y ago
    I guess it's giving you the most likely answer in the same way the most likely answer to the question "what color is the sky?" is blue, according to its training data.
  - swatcoder 4y ago
    Yup. I think this is one of the places where the sheen will come off after a while. While temperature can mix things up a little, we’ll quickly start to see how the depth of LLM output is tied to the depth of LLM input — and people’s prompts are not all that varied and deep at the end of the day. We’ll start to notice signatures to each generation of model in its responses to tiringly common prompts — especially in stuff like blog spam, homework essays, email punchups, casual fiction, etc There are certain poles that it gravitates to that end up reading like verbal tics or lazy ideas, and those will be more obvious as we begin to get inundated with generated content. Careful prompt work can evade those poles and tics, but most people won’t have the skill or drive to bother with that.
    - bgroat 4y ago
      >> Careful prompt work can evade those poles and tics, but most people won’t have the skill or drive to bother with that. I absolutely agree with this. Furthermore, I've come to believe that "careful prompt work" doesn't really save much time relative to just writing/coding the output I want. Especially after QA/edits
      - flangola7 4y ago
        > Furthermore, I've come to believe that "careful prompt work" doesn't really save much time relative to just writing/coding the output I want. The real magic happens when you train a better AI using those perfected prompts.
    - dingclancy 4y ago
      There would be parameters to evade the obvious picks. temperature can actually go up to 2 in the API and you can also go use top_p < 1. but for the more out of this world stuff, this is where fine tuning will come in. The main model will always come in as vanilla anyway.
  - madelyn 4y ago
    I think this is where something like the reAct framework would be super helpful. A human dungeon master doesn't pull a random enemy from the aether super well (or at least I don't!!). When I pick monsters I either logically think what makes sense for the area, pick something that's cool or vaguely humorous that I'm feeling the vibe of, or... use a random encounter table. For treasure and rewards... if it's not something hand picked, I'm again getting the loot tables out and rolling. And heck even for gold rewards and things, I'm rolling the dice. After trying basic stuff with langchain, an AI integrated with these types of tools is really promising. I think more systems on top could make it into a passable DM- at least until context becomes an issue.
  - capableweb 4y ago
    If you set it up with some ground rules and rules around it's behavior (or more like guidelines), you can make it be more creative. Something like: "We're gonna start a text RPG game. Every animal we meet should be a made-up animal composed of joining two existing ones, with creative names being made for them. Don't use any usual fantasy tropes. Make encounters have surprising but logical outcomes" and so on.
- serf 4y ago
  > I guess that something that involves "animated violence" like pokemon might be a good idea to get around that. I feel like this is a bigger thing than people realize. Changing creative output to match the proposed ethical boundaries of a corporation that you're beholden to by using their tool seems like such a new leap to hoist onto the public en-masse. I presume somewhere there is an agreement that forbids the use of 'notepad.exe' in matters of anti-state espionage, homicide, whatever; but this is the first time that a tool is actually good at self-checking to make sure that the license /whatever is upheld rigidly -- and I feel as if that has far-reaching effects on not just the creative output but the thinking structures for those that are involved in the use of such tools. If you know that the tool that you're using forbids certain things, you're either going to think about using it for other things or pursue other tools; given that openAI is effectively the only game in town right now they seem to be in a unique position to affect change in the way people are thinking about using these tools. this could be good or bad, I don't really sway one way or another, but it seems important .
  - MagicMoonlight 4y ago
    It doesn’t take a certificate in applied shoulder massage from lancaster technical college to work out that it’s bad for society. We’ve seen how much social media and manipulation has destroyed the fabric of society. Now imagine those same evil people can control all creative output for their own purpose. Every single thing and person you interact with will be whatever they want. When you whatsapp people they might not even be real anymore, but they’ll certainly tell you why you should vote for president zuckerberg to be reelected. They’ll promise that they’ll only use their control to stop nazism or whatever. But the nazis made the same promise about communists.
- squeaky-clean 4y ago
  I had ChatGPT (not gpt4) semi-succesfully play a session of D&D where I played the dungeon master and it had no problems with violence against goblins. It did however constantly try to take over and play as the goblins as well, and I had to keep telling it that I was the DM, not it. It was fun in the "look at what AI Is doing" novelty way. But it was pretty uncreative and generic.
chaosbolt 4y ago
Using an expensive AI model to run a textual version of a game made for a portable console that came out 2 decades ago is some dystopian shit.
- scrollaway 4y ago
  Why?
  - bluefirebrand 4y ago
    I don't know if I agree that it's "dystopian" but I do somewhat feel it's strange to be "wasting" the massive amount of compute that GPT4 has in order to play a text based version of a Pokemon game that ran on a Gameboy Advance decades ago. It's just funny our priorities with this stuff.
    - minimaxir 4y ago
      Nothing in the original thread asserts that this is more than an intellectual curiosity, which is the ethos of Hacker News. GPT-4 is not going to replace emulating GBA games.
    - swatcoder 4y ago
      “Write my civil war essay” “Write a poem about farts” “Write a JavaScript wrapper for this API” 99% of the usage is a “waste”. Who cares. Some of us are poking around to find that other 1% and some of us are just entertaining ourselves.
      - anthk 4y ago
        If you think about it... people has been emulating the GBA and Pokémon itself on book sized machines faster than a Cray back in the day.
    - dangond 4y ago
      Oh it is absolutely a waste of compute practically speaking. I just find it incredibly interesting that it's even possible for GPT to convincingly recreate many aspects of the original game, including the rules about what is and is not possible at various points, and I haven't seen many others exploring in this direction. I'm very curious as to what Pokemon-simulation-adjacent things GPT is capable of.
ghiculescu 4y ago
This is very cool. I imagine the detailed all-text walkthroughs on GameFAQs would be an amazing training resource. Keen to see it play age of empires next.
aaroninsf 4y ago
VC take note: I would pay streaming subscription prices, for 24/7 access to a LLM chatbot DM which correctly and configurably runs a campaign for me [and friends]. Let me pick between styles and tones (some openly modeled on popular podcasts and streaming play), let me tune a bunch of controls, give me a nice UX if you can, but build me a consistent world which spoons out the world-building _and remembers it_ and I am a customer for life. Oh yeah and the price I will pay doubles if this is illustrated throughout with generative art (also tunable wrt style). We can wait for music and speech synthesis and sound effects for v1.5.
- wesleychen 4y ago
  Maybe something like AI Dungeon: https://play.aidungeon.io/main/home https://play.aidungeon.io/main/home ?
- swatcoder 4y ago
  You don’t want a VC involved in this idea. Their job is to chase extreme growth and burn down any investment that doesn’t grow enough along the way. They’d ruin what you want this to be. But you’ll 100% see this starting to mature as homespun Discord bots in the next year or so, and eventually showing up as either Steam games or web apps from bootstrapped or privately-funded companies soon enough to. It’s not a hard thing to get going, and so it’ll just come down to sorting out a business model and then iterating on features. No VC required. Just time, passion, and some supportive fans/users.
- gaogao 4y ago
  > but build me a consistent world which spoons out the world-building _and remembers it_ Yeah, one potential way to do this is to add input / output to a semantic graph store a la WikiData, in addition to the more bog standard text and image outputs. I've been wanting to try this out ever since seeing how good LLMs are a semantic decomposition.
- abacadaba 4y ago
  https://www.youtube.com/watch?v=MkX9gj_GGwc https://www.youtube.com/watch?v=MkX9gj_GGwc QuestGPT?
kizer 4y ago
I can't wait to read about all of the fascinating applications people will conjure up. I've been playing with it GPT-4 through ChatGPT today. I feel like a little kid with the year's hottest toy.
School-Cotton 4y ago
I can’t understand people who are so excited about GPT. It fills me with absolute dread, and if I could vote to make it illegal, I would. In a world where human intellect becomes meaningless, nobody can predict what will happen next, and a lot of the possibilities are not good. Does anyone want to live in a world where all books, art, etc. are produced by machines, where you can’t get a job tinkering with computers anymore because AI does it better, and so on? The optimistic case for AGI is that if nobody had to work they could spend their time on hobbies and creative passions, but what’s the point when an AI can do them better than any human can? Previous machines only replaced humans at tasks that are tedious and boring, whereas the new crop of AIs seems poised to replace us at the very things that make us human. Not to claim that GPT-4 is all the way there but it’s surely a huge step forward, and I wouldn’t be surprised if we live in such a world in the next decade.
- userbinator 4y ago
  There will always be a place for human intellect. Machines will not replace us. I'm not too worried, as long as they are still reliant on us as their ultimate source of truth.
  - School-Cotton 4y ago
    > There will always be a place for human intellect. Machines will not replace us. I hope you’re right, but how do you know?
    - userbinator 4y ago
      My last sentence. Also, there are several billion of us.
  - flangola7 4y ago
    > There will always be a place for human intellect. Machines will not replace us. I have precisely zero confidence of this and anyone that thinks otherwise is very shortsighted or ignorant or simply lacks imagination. Human intellect is limited by many environmental factors. It has to fit through a 10cm canal, it can't use too much power, it must be made from simple and impure combinations of hydrogen, carbon, and oxygen that have poor conductivity, it must dedicate much of its energy to non-reasoning functions like digestion and respiration control. Just as we made machines with many times our physical strength we can make machines with many times our cognitive strength. The brain is ultimately just an arrangement of about 1500 grams of matter. If we find better arrangements and use better matter and assemble larger amounts of it we will have something with much greater cognitive strength.
qprofyeh 4y ago
This isn’t emulation.
- mrosett 4y ago
  It is in the non-technical sense.
totetsu 4y ago
I was just using the dalle experimental build to generate fakeemon designs. Combine these two and you have an infinite generative monster battle game.
usaar333 4y ago
Pokemon Red also seems supported; haven't tried others. What I find particularly amusing is that (not surprisingly) GPT-4 doesn't really enforce rules. If I want a Pikachu as my starter pokemon, I can get one. (That said it does seem to enforce limits on buying items) Edit: Other games seem to work as well. I just started playing a CLI Deus Ex which correctly starts on Liberty Island and roughly has the correct mechanics.
- dangond 4y ago
  I actually enjoyed its loose enforcement of rules. It made it feel a bit more TTRPG-like, where you could lean into your character to attempt to influence results one way or another.
- minimaxir 4y ago
  When the GPT-4 API goes public, you can edit the system prompt such that it will be forced to follow given rules.
richservo 4y ago
I had a session with ChatGPT that I played a DnD campaign with it, but I told it I was the DM and it was the player. It kept trying to inject story...so I had to keep telling it not to.
LZ_Khan 4y ago
This is hardly "emulating" pokemon emerald. This is like if I was on a camping trip and was so bored I asked my friend to narrate whatever he remembered of pokemon emerald to me.
- 79a6ed87 4y ago
  Exactly, and I came to the comments to see if someone was pointing it out. Anyways, the confusion between what's an emulator and what's a simulator is not uncommon.
  - capableweb 4y ago
    Maybe they mean "emulating" as what's known in common parlor, rather than the specific computer science meaning. "Emulating" is a word that existed before computers, meaning just trying to "match" something by imitation. By that definition, it is trying to emulate the experience of Pokemon Emerald, albeit in text and kind of crappy.
hombre_fatal 4y ago
Man, you can do this with any game. Try "Simulate a text interface to the game Mario", "simulate a text interface to the game Halo". I just beat World 1 and it knew the layout of each level including things that are impossible. Me: Grab star and turn invincible. GPT-4: Well, there's no star here. You can either jump on the goomba, move around, or interact with the question-mark block before you. I'm so uncreative that I need other people to tell me what I can even try with ChatGPT because nothing fun occurs to me beyond asking it boring questions. This is blowing my mind. Though I think it's even crazier to just tell it to create a novel MUD for you to play around in. It will design a text interface, rules for its game, enemies, classes to choose from, spells to use. I just played for 30 minutes in a world it invented on the fly. I'd love a cookbook of things to play with.
- norvvryo 4y ago
  Go full circle and ask ChatGPT for suggestions.
- Sardtok 4y ago
  Was it Multi-User, though?
reiichiroh 4y ago
I’ll be glad when these stunts with AI stop being publicized
cyrialize 4y ago
I was asking ChatGPT how I could use LLMs for setting up a text adventure game. Like using libraries to find certain keywords from a sentence. For example, if the user types a "search" keyword or synonyms, then I'll spit out the response to "search". Then, I realized I could just play the text adventure game with ChatGPT. So I did the following: - I said I had 100 health points and 100 magic points - I have two types of attacks. A normal attack does 25pts of damage to health points and has a low chance of missing. A heavy attack does 50 pts of damage to health points and has a medium to high chance of missing. It leaves me open to attacks. - There are three types of enemies: easy, medium, and hard. An easy enemy has 50 health points and does 15 pts of damage. It has a medium to high chance of missing. Similar stats to medium and hard enemies, with increased health points, increased damage, and lower chances of missing. Then I said let's play an encounter with an easy enemy. It was great! It said I was in a forest, and it told me that there was a rusty sword near by to pick up. I had to fight a goblin for an easy enemy. It was very cool. Made me want to do some more world building, feed that into ChatGPT, and then play within that world.

189 points by dangond 4y ago

I played a RPG session with GPT-3 and it was quite fun. The problem was the combat part, as describing violent events is against the TOS (just when me and a party of villagers had prepared an attack against an orc encampment!) I guess that something that involves "animated violence" like pokemon might be a good idea to get around that.

causality0 4y ago
One interesting weak point I've noticed with GPT is that if you ask it to give examples from a category which in the real world has hundreds or thousands of different possibilities, it will give you the same half dozen answers every time. You ask it to give you "opponents" and it will always give you male humans. You ask it to give you "monsters" to kill and it always gives you orcs, giant spiders, skeletons. If a chest is supposed to have "useful items", it's always a healing potion.
- asddubs 4y ago
  I guess it's giving you the most likely answer in the same way the most likely answer to the question "what color is the sky?" is blue, according to its training data.
- swatcoder 4y ago
  Yup. I think this is one of the places where the sheen will come off after a while. While temperature can mix things up a little, we’ll quickly start to see how the depth of LLM output is tied to the depth of LLM input — and people’s prompts are not all that varied and deep at the end of the day. We’ll start to notice signatures to each generation of model in its responses to tiringly common prompts — especially in stuff like blog spam, homework essays, email punchups, casual fiction, etc There are certain poles that it gravitates to that end up reading like verbal tics or lazy ideas, and those will be more obvious as we begin to get inundated with generated content. Careful prompt work can evade those poles and tics, but most people won’t have the skill or drive to bother with that.
  - bgroat 4y ago
    >> Careful prompt work can evade those poles and tics, but most people won’t have the skill or drive to bother with that. I absolutely agree with this. Furthermore, I've come to believe that "careful prompt work" doesn't really save much time relative to just writing/coding the output I want. Especially after QA/edits
    - flangola7 4y ago
      > Furthermore, I've come to believe that "careful prompt work" doesn't really save much time relative to just writing/coding the output I want. The real magic happens when you train a better AI using those perfected prompts.
  - dingclancy 4y ago
    There would be parameters to evade the obvious picks. temperature can actually go up to 2 in the API and you can also go use top_p < 1. but for the more out of this world stuff, this is where fine tuning will come in. The main model will always come in as vanilla anyway.
- madelyn 4y ago
  I think this is where something like the reAct framework would be super helpful. A human dungeon master doesn't pull a random enemy from the aether super well (or at least I don't!!). When I pick monsters I either logically think what makes sense for the area, pick something that's cool or vaguely humorous that I'm feeling the vibe of, or... use a random encounter table. For treasure and rewards... if it's not something hand picked, I'm again getting the loot tables out and rolling. And heck even for gold rewards and things, I'm rolling the dice. After trying basic stuff with langchain, an AI integrated with these types of tools is really promising. I think more systems on top could make it into a passable DM- at least until context becomes an issue.
- capableweb 4y ago
  If you set it up with some ground rules and rules around it's behavior (or more like guidelines), you can make it be more creative. Something like: "We're gonna start a text RPG game. Every animal we meet should be a made-up animal composed of joining two existing ones, with creative names being made for them. Don't use any usual fantasy tropes. Make encounters have surprising but logical outcomes" and so on.
serf 4y ago
> I guess that something that involves "animated violence" like pokemon might be a good idea to get around that. I feel like this is a bigger thing than people realize. Changing creative output to match the proposed ethical boundaries of a corporation that you're beholden to by using their tool seems like such a new leap to hoist onto the public en-masse. I presume somewhere there is an agreement that forbids the use of 'notepad.exe' in matters of anti-state espionage, homicide, whatever; but this is the first time that a tool is actually good at self-checking to make sure that the license /whatever is upheld rigidly -- and I feel as if that has far-reaching effects on not just the creative output but the thinking structures for those that are involved in the use of such tools. If you know that the tool that you're using forbids certain things, you're either going to think about using it for other things or pursue other tools; given that openAI is effectively the only game in town right now they seem to be in a unique position to affect change in the way people are thinking about using these tools. this could be good or bad, I don't really sway one way or another, but it seems important .
- MagicMoonlight 4y ago
  It doesn’t take a certificate in applied shoulder massage from lancaster technical college to work out that it’s bad for society. We’ve seen how much social media and manipulation has destroyed the fabric of society. Now imagine those same evil people can control all creative output for their own purpose. Every single thing and person you interact with will be whatever they want. When you whatsapp people they might not even be real anymore, but they’ll certainly tell you why you should vote for president zuckerberg to be reelected. They’ll promise that they’ll only use their control to stop nazism or whatever. But the nazis made the same promise about communists.
squeaky-clean 4y ago
I had ChatGPT (not gpt4) semi-succesfully play a session of D&D where I played the dungeon master and it had no problems with violence against goblins. It did however constantly try to take over and play as the goblins as well, and I had to keep telling it that I was the DM, not it. It was fun in the "look at what AI Is doing" novelty way. But it was pretty uncreative and generic.

chaosbolt 4y ago

Using an expensive AI model to run a textual version of a game made for a portable console that came out 2 decades ago is some dystopian shit.

scrollaway 4y ago
Why?
- bluefirebrand 4y ago
  I don't know if I agree that it's "dystopian" but I do somewhat feel it's strange to be "wasting" the massive amount of compute that GPT4 has in order to play a text based version of a Pokemon game that ran on a Gameboy Advance decades ago. It's just funny our priorities with this stuff.
  - minimaxir 4y ago
    Nothing in the original thread asserts that this is more than an intellectual curiosity, which is the ethos of Hacker News. GPT-4 is not going to replace emulating GBA games.
  - swatcoder 4y ago
    “Write my civil war essay” “Write a poem about farts” “Write a JavaScript wrapper for this API” 99% of the usage is a “waste”. Who cares. Some of us are poking around to find that other 1% and some of us are just entertaining ourselves.
    - anthk 4y ago
      If you think about it... people has been emulating the GBA and Pokémon itself on book sized machines faster than a Cray back in the day.
  - dangond 4y ago
    Oh it is absolutely a waste of compute practically speaking. I just find it incredibly interesting that it's even possible for GPT to convincingly recreate many aspects of the original game, including the rules about what is and is not possible at various points, and I haven't seen many others exploring in this direction. I'm very curious as to what Pokemon-simulation-adjacent things GPT is capable of.

ghiculescu 4y ago

This is very cool. I imagine the detailed all-text walkthroughs on GameFAQs would be an amazing training resource. Keen to see it play age of empires next.

aaroninsf 4y ago

VC take note: I would pay streaming subscription prices, for 24/7 access to a LLM chatbot DM which correctly and configurably runs a campaign for me [and friends]. Let me pick between styles and tones (some openly modeled on popular podcasts and streaming play), let me tune a bunch of controls, give me a nice UX if you can, but build me a consistent world which spoons out the world-building _and remembers it_ and I am a customer for life. Oh yeah and the price I will pay doubles if this is illustrated throughout with generative art (also tunable wrt style). We can wait for music and speech synthesis and sound effects for v1.5.

wesleychen 4y ago
Maybe something like AI Dungeon: https://play.aidungeon.io/main/home https://play.aidungeon.io/main/home ?
swatcoder 4y ago
You don’t want a VC involved in this idea. Their job is to chase extreme growth and burn down any investment that doesn’t grow enough along the way. They’d ruin what you want this to be. But you’ll 100% see this starting to mature as homespun Discord bots in the next year or so, and eventually showing up as either Steam games or web apps from bootstrapped or privately-funded companies soon enough to. It’s not a hard thing to get going, and so it’ll just come down to sorting out a business model and then iterating on features. No VC required. Just time, passion, and some supportive fans/users.
gaogao 4y ago
> but build me a consistent world which spoons out the world-building _and remembers it_ Yeah, one potential way to do this is to add input / output to a semantic graph store a la WikiData, in addition to the more bog standard text and image outputs. I've been wanting to try this out ever since seeing how good LLMs are a semantic decomposition.
abacadaba 4y ago
https://www.youtube.com/watch?v=MkX9gj_GGwc https://www.youtube.com/watch?v=MkX9gj_GGwc QuestGPT?

kizer 4y ago

I can't wait to read about all of the fascinating applications people will conjure up. I've been playing with it GPT-4 through ChatGPT today. I feel like a little kid with the year's hottest toy.

School-Cotton 4y ago

I can’t understand people who are so excited about GPT. It fills me with absolute dread, and if I could vote to make it illegal, I would. In a world where human intellect becomes meaningless, nobody can predict what will happen next, and a lot of the possibilities are not good. Does anyone want to live in a world where all books, art, etc. are produced by machines, where you can’t get a job tinkering with computers anymore because AI does it better, and so on? The optimistic case for AGI is that if nobody had to work they could spend their time on hobbies and creative passions, but what’s the point when an AI can do them better than any human can? Previous machines only replaced humans at tasks that are tedious and boring, whereas the new crop of AIs seems poised to replace us at the very things that make us human. Not to claim that GPT-4 is all the way there but it’s surely a huge step forward, and I wouldn’t be surprised if we live in such a world in the next decade.

userbinator 4y ago
There will always be a place for human intellect. Machines will not replace us. I'm not too worried, as long as they are still reliant on us as their ultimate source of truth.
- School-Cotton 4y ago
  > There will always be a place for human intellect. Machines will not replace us. I hope you’re right, but how do you know?
  - userbinator 4y ago
    My last sentence. Also, there are several billion of us.
- flangola7 4y ago
  > There will always be a place for human intellect. Machines will not replace us. I have precisely zero confidence of this and anyone that thinks otherwise is very shortsighted or ignorant or simply lacks imagination. Human intellect is limited by many environmental factors. It has to fit through a 10cm canal, it can't use too much power, it must be made from simple and impure combinations of hydrogen, carbon, and oxygen that have poor conductivity, it must dedicate much of its energy to non-reasoning functions like digestion and respiration control. Just as we made machines with many times our physical strength we can make machines with many times our cognitive strength. The brain is ultimately just an arrangement of about 1500 grams of matter. If we find better arrangements and use better matter and assemble larger amounts of it we will have something with much greater cognitive strength.

qprofyeh 4y ago

This isn’t emulation.

mrosett 4y ago
It is in the non-technical sense.

totetsu 4y ago

I was just using the dalle experimental build to generate fakeemon designs. Combine these two and you have an infinite generative monster battle game.

usaar333 4y ago

Pokemon Red also seems supported; haven't tried others. What I find particularly amusing is that (not surprisingly) GPT-4 doesn't really enforce rules. If I want a Pikachu as my starter pokemon, I can get one. (That said it does seem to enforce limits on buying items) Edit: Other games seem to work as well. I just started playing a CLI Deus Ex which correctly starts on Liberty Island and roughly has the correct mechanics.

dangond 4y ago
I actually enjoyed its loose enforcement of rules. It made it feel a bit more TTRPG-like, where you could lean into your character to attempt to influence results one way or another.
minimaxir 4y ago
When the GPT-4 API goes public, you can edit the system prompt such that it will be forced to follow given rules.

richservo 4y ago

I had a session with ChatGPT that I played a DnD campaign with it, but I told it I was the DM and it was the player. It kept trying to inject story...so I had to keep telling it not to.

LZ_Khan 4y ago

This is hardly "emulating" pokemon emerald. This is like if I was on a camping trip and was so bored I asked my friend to narrate whatever he remembered of pokemon emerald to me.

79a6ed87 4y ago
Exactly, and I came to the comments to see if someone was pointing it out. Anyways, the confusion between what's an emulator and what's a simulator is not uncommon.
- capableweb 4y ago
  Maybe they mean "emulating" as what's known in common parlor, rather than the specific computer science meaning. "Emulating" is a word that existed before computers, meaning just trying to "match" something by imitation. By that definition, it is trying to emulate the experience of Pokemon Emerald, albeit in text and kind of crappy.

hombre_fatal 4y ago

Man, you can do this with any game. Try "Simulate a text interface to the game Mario", "simulate a text interface to the game Halo". I just beat World 1 and it knew the layout of each level including things that are impossible. Me: Grab star and turn invincible. GPT-4: Well, there's no star here. You can either jump on the goomba, move around, or interact with the question-mark block before you. I'm so uncreative that I need other people to tell me what I can even try with ChatGPT because nothing fun occurs to me beyond asking it boring questions. This is blowing my mind. Though I think it's even crazier to just tell it to create a novel MUD for you to play around in. It will design a text interface, rules for its game, enemies, classes to choose from, spells to use. I just played for 30 minutes in a world it invented on the fly. I'd love a cookbook of things to play with.

norvvryo 4y ago
Go full circle and ask ChatGPT for suggestions.
Sardtok 4y ago
Was it Multi-User, though?

reiichiroh 4y ago

I’ll be glad when these stunts with AI stop being publicized

cyrialize 4y ago

I was asking ChatGPT how I could use LLMs for setting up a text adventure game. Like using libraries to find certain keywords from a sentence. For example, if the user types a "search" keyword or synonyms, then I'll spit out the response to "search". Then, I realized I could just play the text adventure game with ChatGPT. So I did the following: - I said I had 100 health points and 100 magic points - I have two types of attacks. A normal attack does 25pts of damage to health points and has a low chance of missing. A heavy attack does 50 pts of damage to health points and has a medium to high chance of missing. It leaves me open to attacks. - There are three types of enemies: easy, medium, and hard. An easy enemy has 50 health points and does 15 pts of damage. It has a medium to high chance of missing. Similar stats to medium and hard enemies, with increased health points, increased damage, and lower chances of missing. Then I said let's play an encounter with an easy enemy. It was great! It said I was in a forest, and it told me that there was a rusty sword near by to pick up. I had to fight a goblin for an easy enemy. It was very cool. Made me want to do some more world building, feed that into ChatGPT, and then play within that world.