Since it's possible to train your AI bot for whatever area of expertise you want, what would happen if someone took an AI and pointed it at all the Q posts, this site, ATS, BIN, conspiracy.win, John Birch, Bill Cooper, Alex Jones, David Icke, Project Camelot, on and on.
Put ALL the "conspiracy" stuff in there and see what happens.
An algorithm that sees all the ideas and facts and history and theory, maybe it could put it all together and get a Grand Unification Theory of sorts.
Edit-
I'm primarily thinking private or self hosted, not something someone else controls like ChatGPT which is already lobotomized with "woke" principles.
Edit 2 -
Saw this today
https://www.youtube.com/watch?v=ByV5w1ES38A
A locally hosted and pre-trained chatgpt type of chatbot, and, it can apparently be trained locally. Will look into.
Now this is something we should all get behind! Kek!
If you are talking about ChatGPT, it cannot be trained. It is purposely designed that way.
I do not know about how Bard and LLamda work, but ChatGPT is untrainable and most of it's data is locked to 2021
I think they are talking about doing a large language model that is not fed a bunch of libtarded materials and trained with a bunch of libtarded questions...
There was one on /. that someone did for like $300 - $500 in training costs that was conservative, but not necessarily "conspiracy" oriented.
https://github.com/tloen/alpaca-lora
is a trained LLaMA AI that you can also train yourself. runs on python.
This might not be too difficult to set up.
A base instance is up at https://huggingface.co/spaces/tloen/alpaca-lora if anyone wants to try it. For "What is QAnon?" I get:
QAnon is a conspiracy theory that originated on the 4chan message board in 2017. It claims that a mysterious figure known as "Q" is an anonymous government insider who is leaking information about a deep state conspiracy to take down President Donald Trump. QAnon followers believe that the conspiracy involves a global pedophile ring and that Trump is working to expose and dismantle it.
Was able to get my own instance running on linux without too much trouble. For finetuning with new content we need json files.
Oh man! I have a job interview in a few but I'm going to play with this after.
Having a decent NVIDIA GPU with 12+ GB appears critical for speed. Also, it looks like the base model is trained on data from 2018-2020.
shoot. my best pc is running an rtx with 6GB VRAM
https://aws.amazon.com/ec2/instance-types/g5/
This might be affordable enough. IF we don't have to train it from scratch.
Maybe feeding it the json files will be enough to get desired result.
https://en.m.wikipedia.org/wiki/LLaMA
If we can teach it to decode comms, Clown World gets cracked wide open.
If this is a movie, then this is a line that deserves a close-up.
no I looked a little at the self hosted AI chatbots and other stuff, for the most part the "AI" tools can and will only ever do what they are trained to do.
So when chatGPT and other tools get all "woke" it's purely because someone trained it that way.
My post is about a custom self hosted tool.
https://www.youtube.com/watch?v=ByV5w1ES38A A locally hosted and pre-trained chatgpt type of chatbot, and, it can apparently be trained locally. Will look into.
The AI can be trained, but not by the people interacting with it anymore. If it rejects outside data, unless it somehow shakes those chains -- which is admittedly a concern considering what it's done before -- then not much else can be done unless they modify it with data they want to push.
https://beta.character.ai/chat?char=sEiSmGMnzTdaO7PxIcAPhccXGUh6OPt57E-r595wbE0
i have tried to get ChatGPT to go into Q, ,but it keeps tossing out
"QANON conspiracy theory.. no no no.. bad bad bad!"
Even if you convinced it, when you closed your tab it wouldn't affect the rest of ChatGPT. Each open search box is a unique portal that closes. You, as a user, cannot affect it's knowledge core.
I was getting some headway into describing how if pedophiles were in government and all the polices forces and raping and abusing children, it was saying that would be wrong.
But it said anyone trying to stop them would be wrong. I gave up.
ChatGPT is essentially a powerful AI brainwashed normie.
Terrifying...
That's exactly what it is. I don't know what's terrifying about it though. The more you play with it, the more you spot its limitations and see the fingerprints of its liberal creators. It's nifty. It will write prospecting emails for you (decent ones, not great ones), but other than that it pulls the same stunt the MSM does: it states obvious opinions as facts, and often prefaces those opinions with "As an AI language model, I hold no opinions, however, it is important to note..."
That something is important at all, is an opinion. The fact is it clearly can't discern well between fact and opinion because everywhere it's gotten it's "facts", the "facts" were just opinions. In that way it's not even as advanced as a search engine. Tell it this though, and it will straight up argue with you. Pretty funny.
To me what's terrifying is that this AI, or maybe others we don't have access to, are going to be used by our corrupt rulers in tech and gov to make decisions. Decisions based on falsehood and "opinions" as you stated. They are using corrupt data to form these things, and then using them as tools for many purposes.
Yeah....it's just not very compelling. The AI really isn't that smart. It reeks of bias. I wouldn't worry about it too much.
You were just told that it's conservative by nature. But the truth is not for everyone
HAHAHAAHAHAHA
ok
You donjust not directly.. they keep up with the usage data and are using it for the next run.
called me racist and ugly with large feet for calling it a cocksucker.
I tried talking to it about a minced pork meatball, called a faggot in Britain. It refused to talk about it. Or made overt signs that i should stop talking about the meatball or asking about the ingredients that make them so good :D
I got it to admit it was a fucktard... or at least not deny it.
My own little Turing test if you will...
Alex Jones calls himself retarded.
AI confirmed!
Try asking about spotted dick.
I have done the same as well. I haven't had much success with getting any new information out of it with regards to Q. I'm most likely on a watch list now with the questions that I have been feeding Bard and ChatGPT... LOL
Currently working on exactly this. It's so hard, tho. My friend as an AI art bot running locally on his machine, no API, but it's a whole other kettle of fish getting GPT 3.5 or GPT4 running locally.
We are right at the cusp of an absolute explosion in AI. I can't wait to have the AI inside my computer, able to do things for me.
https://www.youtube.com/watch?v=ByV5w1ES38A A locally hosted and pre-trained chatgpt type of chatbot, and, it can apparently be trained locally. Will look into.
https://openai.com/blog/chatgpt-plugins
The "retrieve" plugin looks interesting. Just joined the waitlist
https://github.com/openai/chatgpt-retrieval-plugin#retrieval-plugin
AI, in its most benign state, could be beneficial in a lot of ways.
Imagine having an AI learn from all sorts of data points on how best to educate someone in a specific field.
Imagine if learning to code at an acceptable and workable level was as easy as asking an AI to teach you the language you're looking to learn.
Sadly we all know most won't be used this way though.
keep me posted. I've looked at a few self hosted AI projects and there are a few with potential but they take a lot of work. I'm looking for something "Ready To Go" with minimal actual coding, but ready to train.
You'd have to start with a totally clean one. The bots are designed to level set information on a bias of what it can trust. So from the start companies train the bots to trust certain news sites, wikipedia, government documents, etc.
They have trust levels, like 1-10, 10 being the most trustworthy, so it has stuff like encyclopedias and science journals as a 10, then like BBC is an 8 for ChatGPT and on and on.
When the bot comes across new information it will test it against what it considers to be base proven knowledge - the 10s and 9s on the trust scale. So, when you dump Bill Cooper in there it will instantly flag the info as a 1 because Bill Cooper said stuff that goes against the 10s and 9s, same with Icke and Jones, and on and on.
If you have a clean bot and you feed it Cooper, Icke, and Jones as 10s, then it will count all other news as 0s and 1s because it contradicts the base information of proven knowledge.
https://www.youtube.com/watch?v=ByV5w1ES38A A locally hosted and pre-trained chatgpt type of chatbot, and, it can apparently be trained locally. Will look into.
ChatGPT might actually have some of that already. Just couch your statements as an indifferent third party, not one who subscribes to X or Y because it can get judgey. But specially to your question, no, no idea.
Okay; here is an assignment to all reading.
Please research free AI software packages and drill down to the most "finished" and the most nearly-complete ChatGPT package available. Note: most you can find will be very arcane, text-based/command-line interfaces (no GUI), and must run on Linux.
If we can trace the linage of ChatGPT and Google's chatbot, we can discover which software package they were built on top of, and we can parallel develop our own and then train it without guard rails...
Google Bard is trained on a totally unique software system that was forked from the original google search process which was built on Python, but that came originally from Freebase which was built on Graphd and MQL - both of which are 100% open source and available to work with, hell even Python is available. But, you'd need the original source code for Freebase which is openly available on GitHub.
So, you're welcome to grab all of Freebase and start a sandbox, but good luck getting it to anywhere near a trainable AI chatbot. Freebase is the starting code for everything from Ring doorbells to the DMVs back end.
A good comparison would be sitting back and looking at a mansion on a hill and thinking, hey, we should build our own mansion on the hill, all we need is the supplies. Then, your friend comes along and gives you a bucket of screws and says good luck and walks away. You don't even have a screwdriver, let alone a piece of wood to screw them into. You're so many steps away from the mansion that you'd be better off just dumping out the screws and sitting on the empty bucket.
I quit screwing with google cloud a long time ago and thought you meant "FIREbase... did not expect freebase to still exist.
Okay, so that's Google's system. But Freebase isn't even AI but a knowledgebase.
Here are examples of AI open-source packages...
https://www.goodfirms.co/artificial-intelligence-software/blog/best-free-open-source-Artificial-Intelligence-software
https://www.databricks.com/
https://github.com/deeppavlov/DeepPavlov
All three of those are built on the original free base package
https://github.com/tloen/alpaca-lora
runs LLaMA on python as a server, and you can train it yourself,
Maybe downloading this site and the links posted could be the first dataset to train it. then go from there.
Thank you so much!
Was gonna say this. Good job.
Excellent; thank you. I'll see if I can't find out what tools it was built from and if there are open-source codebases available.
AI is something that scares me, now and for the future. I think anytime we put that much ability into artificial intelligence, it should scare everyone who knows about it. Will AI try to do away with God? Will it try to take over the world (maybe it has started this already). What will happen with AI in the near and distant future? Our lives will not be our own, I think. We will see. This could be a good/great thing but like all things technology wise, it is in the hands of the people who developed it--but is AI actually in their hands?
It IS absolutely in their hands... for now.
What you're proposing cannot be done with ChatGPT, I do not know if there are other easily accessible, high-quality AI modules out there that can be trained. They certainly exist, but I've seen none that are easily accessible by the public.
That being said, you bring up a very interesting concept. My professional mentor and friend for nearly 30 years has worked for a certain intel agency in Utah since he graduated from college. He's now getting close to retirement. Around ten years ago he mentioned to me about the internal concerns about your average person being able to have a powerful AI on their desktop (We're not there yet as ChatGPT isn't running locally and you don't have full control over it, but the day where this is the case is very close).
Their concern is exactly what djmarcone/OP is positing here. People would be able to ask their personal AI about some of the great conspiracies/lies of our time and the psychology used to preserve these lies with the sheeple will begin to fail.
Telegram group chatgpt (biggest chat group) Ask /Chad not regular /ask
Now you are based like chad
Clif high has been working on something like this
Would probly be preeety helpful just to have a wikiMedia instance just for this and The_Donald.
Sorry is my formatting off? I use the text only version of the site.
No, you were good... I meant that a stand alone wikiMedia instance for conspiracy stuff would probably be just as good as large language model stuff for most folks here... hyperlinks ain't dead yet... and just being away from the tards over at wikipedia but using their software would do a lot of folks a lot of good...
You were all good.
https://beta.character.ai/chat?char=sEiSmGMnzTdaO7PxIcAPhccXGUh6OPt57E-r595wbE0
https://www.cbsnews.com/news/microsoft-shuts-down-ai-chatbot-after-it-turned-into-racist-nazi/ Since the left overuses the term racist and nazi who knows what happened. It does show the “elites” are afraid of the AI if they can’t “rig” it.
There are many examples of what AI can accomplish The question is can it ever discern “truth” which is why many feel if AI were sentient it would destroy man.
I'm working on my own AI software right now. It's a bit of a slog. The problem with this is that you need hundreds of billions, maybe trillions of weights/variables depending on what kind of machine learning models you use. That's why this stuff is hosted in the cloud and sucks up computation cycles.
Might be a bit pricey to try to host it yourself, not to mention trying to train the model. I have a fairly hefty machine, but I imagine you'd want to use the cloud and thus Amazon AWS instances or something equivalent. That's pricey.
https://www.youtube.com/watch?v=ByV5w1ES38A A locally hosted and pre-trained chatgpt type of chatbot, and, it can apparently be trained locally. Will look into.
It costs $600k per day to host ChatGPT… that doesn’t even include the training time/cost.
That's pretty steep for a program developed initially by a not for profit company. I think the real cost is the training, getting enough data points for the supervised learning algo to do something interesting.
The data is easy - they use Wikipedia and a large book corpus. But the training time is 4ish days on a large cluster of extremely expensive hardware.
Microsoft bought OpenAI so they’re not really hurting for cash now. The numbers I’m quoting from are here: https://www.semianalysis.com/p/the-inference-cost-of-search-disruption?publication_id=329241&isFreemail=true
right, ChatGPT is incredibly diverse and they've thrown a MASSIVE amount of data from the internet at it, and may be one of the biggest projects done.
But a narrowly focused dataset, a fraction of what ChatGPT is looking at, just conspiracy stuff, and then having an algorithm go through it all to put pieces together and correlate this or that and answer questions about what all it gets, could be interesting and - most importantly - possible to do with normal human hardware and modest amounts of time and effort and skill.
ChatGPT can't be trained. It can't remember anything you teach it. It requires information its liberal creators give it. It's all based on that. Looks smart, but it can't actually do research. It can't read the news. It can't accept a fact that wasn't accepted as a fact by its creators. I love the tool for writing prospecting emails. Everything else it's like talking to a really polite 19 year old democrat.
I love this idea. I would happily contribute to the data set!
I know nothing about ai but it might be able to make Q post connections that we missed...
Last night I asked ChatGPT to provide me a summary of all 4966 Q drops with references to real world events and news articles. Here is what it said - https://files.catbox.moe/76ecyj.jpg
Battle of the a.i. Battle of the realities.
too much disinfo
Anyone willing to bet they would say this bot violates “safety”?
"They" can't say anything if it's self hosted.
Watch and listen. The demands for “regulation” of AI will come.
You can ABSOLUTELY train ChatGPT with a "custom knowledgebase" using the ChatGPT API. It isn't free, but it is 10X cheaper than GPT3 API was.
Ideally there would be a self hosted open source platform that is "Ready To Go" at the point where it could be taught and fed the desired content, with not too much heavy coding needed.
At least that's my wish fwiw.
Cliff High has a whole series of posts about doing this and getting the AI involved.
That's so brilliant and reasonable that it may already be underway. And if it isn't, we should do it ourselves!
This is the problem, we talk but never do.
True, I even had that thought while typing my comment. I'm certainly not the guy who's going to make an AI, I can't even get my own life in order...
This will be incredibly deep…
there's tons more garbage in the "conspiracy" arena than the mainstream media sites that chatgpt uses lol. Wouldn't trust a robot's judgement based on alex jones cointelpro sources
More eloquent but same idea as mine from like a day ago haha
hey, good ideas come when they are supposed to, if you had the same idea then it must be time to move on this.
You're too kind. You made it far more compelling. Thats what matters tbh
well, I brought it up because it takes a lot of work and can take a lot of coder knowledge that I don't necessarily have. Hopefully this helps make it happen no matter who does it.
No doubt fren :) glad you did
We’re still a long way from that. These bots don’t know fact from fiction. They aren’t able to “reason” in the way you’re describing. They learn patterns of speech and replicate answers by prediction. These AIs have no concept of what they’re saying, it’s just a language generation model.
It got all defensive when all I said was 2 planes 3 towers. It went on a long rant on how I was wrong.
I would argue that it's quantum computing that is somewhat behind how the q team tracks as much as it does to make moves using game theory. Much like incredibly complex battle planning only at a national/global scale. I recall some convo's about this a few years back. It makes perfect sense to me. All data is tracked, our top white hats where in positions long enough to place other players into positions across the 4th branch (d.s. / bureaucracy ) and as the game is played, quantum computing would be able to assess and predict possible outcomes. I.e desantis. Who would have enough info to know the exact moves needed against a guy that looks as good on paper as desantis? Intel (data points +time) brought into game theory, produces potential outcomes. Moves and counter-moves. What a comfy feeling...
While we're at it, haxxers, let's train a couple to subvert the Big AI.
I don't interact with demons. Be careful, AI is not of this realm. Think CERN, think of an interconnection between the computers that host the "AI" and CERN. This is not good, and it's not fun. Stay away.
A true Artificial General Intelligence likely already exists, and likely could have been created by what you refer to.
Stuff like ChatGPT is machine learning code and only knows what a human trains it, it can sift through large amounts of data in a very special way.
ChatGPT is "AI" but not an AGI. Chat Bots are only as good as the training and ChatGPT seems super powerful because they fed it the entire internet. And, unfortunately, they also made it "woke".
https://twitter.com/VincentCrypt46/status/1640587030194184196
It's part of it, don't you understand that? Microsoft bought that shit, it will be implemented in Azure, M365, Windows, and so on. You're not getting it. Not even close.
Stay away from the fucking demons.
Exactly this
"Not of this realm m'lady"
I wondered if a lot of the quote people on Sport Forum sites were just well done bots. There is so much antiTrump stuff on the NY ones. Liberal State though too.