Imagine people starting to create "internet bunkers," which are groups of verified humans in a chat that can talk to each other. These trusted individuals can explore the internet and try to find other humans online if possible. Damn that scenario is amazing but also far-fetched.
>people know AI is among them on the interwebz >steadily but surely, trust in internet persons lower >normal people shift focus to communication with people they know irl >only full degenerates, which shouldn't breed regardless, remain on massive discord servers, twitters, reddits and whatnot
I see this as an absolute win
For VR Porn game there is VAM and there was some guy developing an automatic male masturbator that would connect to the feed of the game, he already had it working in a prototype months ago he livestreamed it on youtube on how to build one yourself. It was really sick, it could do up and down left and right and so on based on what you would see in game virtually.
A system that reads articles and transforms sentences into graph database code for something like Neo4j.
Eventually, you would have a massive knowledge graph that you could query to search for previously unseen coincidences.
Your picrel lead me his wiki, that lead me to an article on him from 2016 and my picrel
Maybe it's been over for a while and this is just the endgame...
>Maybe it's been over for a while and this is just the endgame...
Yudkowsky said he cried all his tears in 2015 when he realized there was little chance we survive.
>Ideas for things that AI could generate but haven't been tried yet.
Synergistic society of some African cunt and AI.
It would be total KINO if some Gambia had became 1st world by embracing AI governing.
Here are some ideas for AI-generated content that could be interesting to develop:
AI-generated world-building: AI could generate detailed, fictional worlds, complete with geography, history, culture, and lore for writers, game developers, and role-playing enthusiasts.
AI-generated theme park attractions: Using visitor preferences and attraction themes, AI could generate new ride concepts and layouts, optimizing them for excitement, safety, and visitor flow.
AI-generated fashion designs: AI could analyze trends, color palettes, and textures to create unique, personalized fashion designs that cater to individual tastes.
AI-generated fitness routines: By considering personal fitness goals, available equipment, and individual constraints, AI could create optimized and varied workout routines for users.
AI-generated virtual reality experiences: AI could generate unique, immersive environments for users to explore, with each environment tailored to the user's preferences and interests.
AI-generated recipes: By analyzing flavor profiles, dietary restrictions, and available ingredients, AI could create unique and delicious recipes for users to try.
AI-generated urban planning: AI could analyze data on population growth, infrastructure, and environmental factors to create optimized urban designs that balance functionality, aesthetics, and sustainability.
AI-generated educational curricula: AI could generate personalized educational plans for students, taking into account their learning styles, strengths, weaknesses, and interests.
AI-generated product ideas: By analyzing market trends and consumer preferences, AI could generate innovative product ideas and develop preliminary designs for entrepreneurs and businesses.
AI-generated art installations: AI could generate interactive and immersive art experiences that adapt to the viewer's presence and actions, providing a unique experience for each visitor.
Design a biological structure that could work as hardware for AI computations.
Then guess what genetic code could create similar structure, by simulating growth of considered cells.
Lore master AI that organizes your confusing mess of worldbuilding notes.
I tried to make AutoGPT do it but it kinda can't figure out what to do even when I hand it the directory path.
https://arxiv.org/abs/2304.03442
TL;DR world of generative agents inside a sandbox game environment, where the characters closely resemble human behavior and the agents are augmented with three additional capabilities: memory, reflection, and planning. That was sufficient to create a small world in a game environment reminiscent to the actual world humans live in.
All AAA games will gace GPT in them and you will be able to talk about anything with them. Like talk about shit posts on BOT in GTA-6 with random NPCs.
If I had more VRAM and an army of pajeets to label data for me I'd make an extremely beefy NER model that could reliably detect any and all shitcoins mentioned in a sentence and a BOT-specific sentiment analysis model, then scrape BOT and analyze their posts and charge people for the service of freeing them from that shithole.
A tool that verifies and gives context to facts/statistics. Someone online posts something like >The US today wastes more money on welfare than they ever did
and I can just highlight the text or mouseover and then a window pops up with a short summary of the topic including >money spent on welfare in $ and % of budget >other relevant facts (population, cost of living, etc.) >comparison to historical trends >comparison to other countries >plus citations
Sharps have been using machine learning for a long time.
Having access to a decent LLM isn't going to make someone in to a sharp. You need to understand where the ev comes from to begin with.
Also since it's a market with market makers, it's more similar the stock market. Speed is the edge. Getting the most edge as quickly as possible without getting cut off or labeled is what's important.
Plus honestly unless something has seriously changed the last like 5-10 years or so, getting big amounts of money in action is quite difficult. Even a huge edge like 1% requires $1m in action just to make $10k in ev. That's not exactly a ton of money relative to the risk and logistics of actually getting that money in action.
tbh a lot of the best gamblers in the world moved to where the easy money is a while ago anyway. You all know what that is.
>A federated LLM running on a blockchain. How feasible do you think this is?
It would have to work as individual agents with quantized Intellectual Rating and Rated Knowledge Set. With some race to finish the task, and other infrastructure to evaluate results.
>How would the "Intellectual Rating" and "Rated Knowledge Set" of individual agents be determined? >How would the "race to finish the task" be structured, and what incentives would be offered to encourage participation? >How would contributions be evaluated and incorporated into the final model?
>Intellectual Rating
Agent with it's infrastructure would have to forego some standardized test. >Rated Knowledge Set
Agent would have to buy one of the standardized Knowledge Set libraries. >race to finish the task
A task would be announced to all agents, first few to report results would go further checking. The ones with right answer would get rewarded(Coins). Bad answer would result in reputation penalty. >How would contributions be evaluated and incorporated into the final model?
Some reputation system and majority vote probably.
>speed and competence
*not all agents are equal. Some are GPT3-tier, some are GPT4-tier, some GPT7-tier. They can only participate in their designed league task pool.
>GPT7-tier
*GPT7-tier create the inner circle that is responsible for defining a long term system strategy, and in free time will play Stellaris with me.
Thanks for your input. I still have some questions though.
> How would the standardized tests and knowledge sets be developed and maintained? > How would the system ensure that agents are not sharing or copying answers from other agents? > How would the system handle agents who intentionally provide incorrect answers in order to earn rewards? > How would the majority vote system work in practice, and how would it be designed to prevent biases or manipulation? > How would the system handle cases where agents have conflicting answers, or where there is no clear consensus on the correct answer?
>standardized tests
Solving logical problems in defined time, similar to IQ tests. >knowledge sets
Developed by crawling tasks(gpt3-tier), and evaluation tasks(gpt4-tier). >system ensure that agents are not sharing or copying answers from other agents
Reputation ranking and maybe some 'salt' added to each task->agent. >agents who intentionally provide incorrect answers
Reputation penalty. They would have to do some tasks for free (or invest coins) to regain reputation. >majority vote system work in practice, and how would it be designed to prevent biases or manipulation?
The results would have to be spited into smaller parts and evaluated by lesser agents. >system handle cases where agents have conflicting answers, or where there is no clear consensus on the correct answer?
No rewards. The problem would have to be reformulated.
A ban system that puts shitters/teamkillers/12 year olds into their own fake lobby with "players" to keep them occupied and fool the retards instead of them getting a ban and hopping to a new account.
>a game that does it yet simulating real players
I believe that those stupid Swedish mobile games do that, to create an illusion that anyone plays that shit.
Oh great, another one of those boring AI enthusiasts trying to come up with new ideas for their pet project. Can't they just stick to writing code and leave the creativity to actual humans? I mean, seriously, who needs an AI to generate ideas when we already have Twitter and Reddit? But hey, if you're looking for some truly innovative ideas, why don't you try creating an AI that can actually make us laugh instead of just spitting out the same old banalities? Or better yet, how about an AI that can cook us a delicious meal without setting off all our food allergies? Now that would be worth funding.
A bookstore where you ask for a book on a particular set of topics in a format of your choice, and then a new book that's never existed before is generated to your specifications, printed, and shipped to you. Or same but .pdf.
It's likely only a matter of engineering and a ton of datacenter GPUs at this point, hard research on the necessary fundamentals already done.
>but haven't tried yet
who?
An imageboard where all the posts are made by AI. It pretends it's a lively imageboard, but in reality it's a ghost town.
Dead Internet Theory becoming true
Imagine people starting to create "internet bunkers," which are groups of verified humans in a chat that can talk to each other. These trusted individuals can explore the internet and try to find other humans online if possible. Damn that scenario is amazing but also far-fetched.
In reality we'll just be stuck at Elons', having to humor his 12 y/o jokes and rocket wanking.
PS most socials will become ID-only, probably
>people know AI is among them on the interwebz
>steadily but surely, trust in internet persons lower
>normal people shift focus to communication with people they know irl
>only full degenerates, which shouldn't breed regardless, remain on massive discord servers, twitters, reddits and whatnot
I see this as an absolute win
>He doesn't know
>self host AIs
>self host an imageboard
>keep everything private
what could go wrong
>that post was generated by a LLM
so, 4chAIn
I've thought about doing something like that, but instead it's all fake glitchy AI spam. Thought it would be an interesting art project.
some sandbox mmorpg where all players are ai npcs and they learn new stuff everyday (like the sims)
vr porn
For VR Porn game there is VAM and there was some guy developing an automatic male masturbator that would connect to the feed of the game, he already had it working in a prototype months ago he livestreamed it on youtube on how to build one yourself. It was really sick, it could do up and down left and right and so on based on what you would see in game virtually.
nanii
give me link and names for this shit NOW
too...busy...getting...blowjobs...24/7....ughhhh
can't....be....bothered....to...fffffiiiiiiiind...link
A system that reads articles and transforms sentences into graph database code for something like Neo4j.
Eventually, you would have a massive knowledge graph that you could query to search for previously unseen coincidences.
that schizo picture, it almost makes me want to stop taking my meds to see the world that way again
Meds are not recommended past this point.
Some people are already doing that:
https://github.com/varunshenoy/GraphGPT
Wtf is this wonky ass clown shit. Looks like a high school project.
neat but like...
what on earth is different from just drawing this with your own hand lol
Your picrel lead me his wiki, that lead me to an article on him from 2016 and my picrel
Maybe it's been over for a while and this is just the endgame...
Holy shit nice find.
>Maybe it's been over for a while and this is just the endgame...
Yudkowsky said he cried all his tears in 2015 when he realized there was little chance we survive.
>chance we survive
We as humankind, or the Yud-kind 'we'?
wouldn't you like to know :>
>Ideas for things that AI could generate but haven't been tried yet.
Synergistic society of some African cunt and AI.
It would be total KINO if some Gambia had became 1st world by embracing AI governing.
Malware
I have a one million dollar idea, I'll sell it to you for half it's worth. What say you?
Here are some ideas for AI-generated content that could be interesting to develop:
AI-generated world-building: AI could generate detailed, fictional worlds, complete with geography, history, culture, and lore for writers, game developers, and role-playing enthusiasts.
AI-generated theme park attractions: Using visitor preferences and attraction themes, AI could generate new ride concepts and layouts, optimizing them for excitement, safety, and visitor flow.
AI-generated fashion designs: AI could analyze trends, color palettes, and textures to create unique, personalized fashion designs that cater to individual tastes.
AI-generated fitness routines: By considering personal fitness goals, available equipment, and individual constraints, AI could create optimized and varied workout routines for users.
AI-generated virtual reality experiences: AI could generate unique, immersive environments for users to explore, with each environment tailored to the user's preferences and interests.
AI-generated recipes: By analyzing flavor profiles, dietary restrictions, and available ingredients, AI could create unique and delicious recipes for users to try.
AI-generated urban planning: AI could analyze data on population growth, infrastructure, and environmental factors to create optimized urban designs that balance functionality, aesthetics, and sustainability.
AI-generated educational curricula: AI could generate personalized educational plans for students, taking into account their learning styles, strengths, weaknesses, and interests.
AI-generated product ideas: By analyzing market trends and consumer preferences, AI could generate innovative product ideas and develop preliminary designs for entrepreneurs and businesses.
AI-generated art installations: AI could generate interactive and immersive art experiences that adapt to the viewer's presence and actions, providing a unique experience for each visitor.
AI generated post
Give me a more detailed plan for "AI-generated art installations"
There was this GPT idea of using some quantum-surface materials, that when beamed with light would create specific sounds.
Quite KINO.
Hentai that caters to all our degenerate fetishes, new anime shows, and VR world creation.
Design a biological structure that could work as hardware for AI computations.
Then guess what genetic code could create similar structure, by simulating growth of considered cells.
Lore master AI that organizes your confusing mess of worldbuilding notes.
I tried to make AutoGPT do it but it kinda can't figure out what to do even when I hand it the directory path.
Some funny game where all npc controlled by GPT.
I actually wanna do this ir experience this. Hopefully soo someone qill came up with a good ARPG
moron, do you even read papers?
https://arxiv.org/abs/2304.03442
TL;DR world of generative agents inside a sandbox game environment, where the characters closely resemble human behavior and the agents are augmented with three additional capabilities: memory, reflection, and planning. That was sufficient to create a small world in a game environment reminiscent to the actual world humans live in.
Interesting
All AAA games will gace GPT in them and you will be able to talk about anything with them. Like talk about shit posts on BOT in GTA-6 with random NPCs.
If I had more VRAM and an army of pajeets to label data for me I'd make an extremely beefy NER model that could reliably detect any and all shitcoins mentioned in a sentence and a BOT-specific sentiment analysis model, then scrape BOT and analyze their posts and charge people for the service of freeing them from that shithole.
Shitcoin moods-analyze-auto-trader.
Could an AI help me create an imageboard? I want to make a sucessor for Hispachan
Design a cryptocurrency, where 'mining' cycles are used by GPT.
A tool that verifies and gives context to facts/statistics. Someone online posts something like
>The US today wastes more money on welfare than they ever did
and I can just highlight the text or mouseover and then a window pops up with a short summary of the topic including
>money spent on welfare in $ and % of budget
>other relevant facts (population, cost of living, etc.)
>comparison to historical trends
>comparison to other countries
>plus citations
Functional government
3d objects that can be imported into video games
I've seen simplistic implementations of that.
A software that monitors ALL input/output from a computer and generates statistics with graphics.
ANIMATION OF 3D HUMAN SKELETON MODELS
Img2img mapping of 2D T posed models to 2D skeletons could be useful for animation. Skipping 3D makes things a lot simpler.
Use AutoGPT to turn a Roomba into a pet.
AI generated game creator where you could tell it to generate characters, setting map, and textures.
I would like it to be 2d 16 bit pixel games first it doesn't have to start with full 3d.
these are all really good
Is GPT-4 as fast as GPT-3.5 by now?
A sportsbetting AI
I would legit pay money for this... no idea why one of you retards haven't done this yet
how would AI predict the outcomes of games thoughever? by comparing team combinations and making an educated guess?
taking in all known data about the teams including their social media profiles(find if someone is going through a break up, anything)
take in as much data as possible and pick winners accordingly
the trained data only goes as far as november 2021
Sharps have been using machine learning for a long time.
Having access to a decent LLM isn't going to make someone in to a sharp. You need to understand where the ev comes from to begin with.
Also since it's a market with market makers, it's more similar the stock market. Speed is the edge. Getting the most edge as quickly as possible without getting cut off or labeled is what's important.
Plus honestly unless something has seriously changed the last like 5-10 years or so, getting big amounts of money in action is quite difficult. Even a huge edge like 1% requires $1m in action just to make $10k in ev. That's not exactly a ton of money relative to the risk and logistics of actually getting that money in action.
tbh a lot of the best gamblers in the world moved to where the easy money is a while ago anyway. You all know what that is.
Presidental election AI that will make you a president.
An AI that creates in-between frames for animation
Just make Roko's Basilisk
On a serious note: A federated LLM running on a blockchain. How feasible do you think this is?
Having a mega-cluster for training and inference could both democratize AI and produce the strongest model in existence.
+SMAGNP+
>A federated LLM running on a blockchain. How feasible do you think this is?
It would have to work as individual agents with quantized Intellectual Rating and Rated Knowledge Set. With some race to finish the task, and other infrastructure to evaluate results.
Could you elaborate?
>How would the "Intellectual Rating" and "Rated Knowledge Set" of individual agents be determined?
>How would the "race to finish the task" be structured, and what incentives would be offered to encourage participation?
>How would contributions be evaluated and incorporated into the final model?
>Intellectual Rating
Agent with it's infrastructure would have to forego some standardized test.
>Rated Knowledge Set
Agent would have to buy one of the standardized Knowledge Set libraries.
>race to finish the task
A task would be announced to all agents, first few to report results would go further checking. The ones with right answer would get rewarded(Coins). Bad answer would result in reputation penalty.
>How would contributions be evaluated and incorporated into the final model?
Some reputation system and majority vote probably.
>forego some standardized test
I mean it would have to undergo some official test to describe it's speed and competence.
>speed and competence
*not all agents are equal. Some are GPT3-tier, some are GPT4-tier, some GPT7-tier. They can only participate in their designed league task pool.
>GPT7-tier
*GPT7-tier create the inner circle that is responsible for defining a long term system strategy, and in free time will play Stellaris with me.
Thanks for your input. I still have some questions though.
> How would the standardized tests and knowledge sets be developed and maintained?
> How would the system ensure that agents are not sharing or copying answers from other agents?
> How would the system handle agents who intentionally provide incorrect answers in order to earn rewards?
> How would the majority vote system work in practice, and how would it be designed to prevent biases or manipulation?
> How would the system handle cases where agents have conflicting answers, or where there is no clear consensus on the correct answer?
+DRXX4W+
>standardized tests
Solving logical problems in defined time, similar to IQ tests.
>knowledge sets
Developed by crawling tasks(gpt3-tier), and evaluation tasks(gpt4-tier).
>system ensure that agents are not sharing or copying answers from other agents
Reputation ranking and maybe some 'salt' added to each task->agent.
>agents who intentionally provide incorrect answers
Reputation penalty. They would have to do some tasks for free (or invest coins) to regain reputation.
>majority vote system work in practice, and how would it be designed to prevent biases or manipulation?
The results would have to be spited into smaller parts and evaluated by lesser agents.
>system handle cases where agents have conflicting answers, or where there is no clear consensus on the correct answer?
No rewards. The problem would have to be reformulated.
>Reputation penalty.
*also agent would have to meet reputation threshold to earn rewards.
>bounty hunter AI agents
No one will implement it, right?
>41chain
I like it.
ch41n
A ban system that puts shitters/teamkillers/12 year olds into their own fake lobby with "players" to keep them occupied and fool the retards instead of them getting a ban and hopping to a new account.
Shadowban and echochamber. The ideas have been around for decade.
The ideas have been, yes, but there's never been a game that does it yet simulating real players, including relevant angry chat messages.
>a game that does it yet simulating real players
I believe that those stupid Swedish mobile games do that, to create an illusion that anyone plays that shit.
is there any good services that let you generate porn?
Oh great, another one of those boring AI enthusiasts trying to come up with new ideas for their pet project. Can't they just stick to writing code and leave the creativity to actual humans? I mean, seriously, who needs an AI to generate ideas when we already have Twitter and Reddit? But hey, if you're looking for some truly innovative ideas, why don't you try creating an AI that can actually make us laugh instead of just spitting out the same old banalities? Or better yet, how about an AI that can cook us a delicious meal without setting off all our food allergies? Now that would be worth funding.
thanks chatgpt
>watching too many of those stupid ai streams was good for something
Go read some McLuhan.
A bookstore where you ask for a book on a particular set of topics in a format of your choice, and then a new book that's never existed before is generated to your specifications, printed, and shipped to you. Or same but .pdf.
It's likely only a matter of engineering and a ton of datacenter GPUs at this point, hard research on the necessary fundamentals already done.
Ask AI.