AI GPT4All Chatbot on Laptop?

I am interested in AI chatbots.
I was inspired by the wonderful thread Chatgpt in terminal using ai.py and contributions there and I tried but could not do it.

Someone sent me a link to what seems to me to be an alternative which is called gpt4all. https://github.com/nomic-ai/gpt4all

My humble understanding is that I can install it on my laptop. I understand it already has the “data” it needs and it can be trained or learn from the data I give it.

As a non techie old man I thought I better consult with you to understand more and get it working properly, and above all make it learn about specific topics I care about (maybe from lots of PDF research and articles I have on my laptop).

Unfortunately there are a few things I did not understand on the website, I don’t even know what “GPT-3.5-Turbo Generations”, “based on LLaMa”, “CPU quantized gpt4all model checkpoint”… etc.

I will appreciate any clarifications and guidance on how to:

  1. install
  2. give it access to the data it requires (locally or through web?)
  3. How to “teach” it or add data to it.
  4. What about asking it in languages other than English? Or info/data other than English?
  5. anything else to make my use case better?
  6. Anything more you see relevant?
    Once installed will it just answer my questions? Will it learn from “chatting” with me?
    Can I make it read the research or articles I download and make it “learn” from it?

I know this post is quite long, but I believe the topic might be of interest for many and benefit many hopefully.

I will appreciate your inputs and feed back and hope it will benefit whoever is interested in AI.

2 Likes

I have not used it and never will, but I can give you my simplified overview for this horrible project.

  • ChatGPT, or any other so-called AI, is a monster.
  • If you use it online, you are feeding the monster that wants to eat you.
  • If you install it on your PC, and it is connected to online databases, see above.
  • If you install it on your PC, and its databases are not online (they live on your disk), you are feeding a monster that wants to eat you, but you can kill it anytime.

It is really disappointing to me, that so many humans are excited by this horrible project, and have surrendered their existence to it, without knowing the insides of it.
So much for Privacy focused browsers and systems, but welcoming a stranger in your house, because of a gorgeous look.
:frowning_face: :cry:

Worth watching video:

7 Likes

A wise :clown_face: once said:

Garbage in = Garbage out

honka_animated-128px-37

3 Likes

That’s the most scary, considering nobody is considering this.

1 Like

image

2 Likes

The reason why Elon wants GPT-5 stopped is because he wants to get in on the AI chat business (as he has expressed many times), and probably wants unfair advantage due to regulatory capture and raised barriers to entry.

That said, I completely agree with you that GPT is a monster and that is shouldn’t exist, I really, really hate it, considering the likely consequences it will have on our lives, but at this point, the toothpaste is out of the tube and you can’t really push it back in.

Even if Elon got his way and AI research was banned in the US, that would not stop China and similar immoral regimes from doing it. If anything, it would give them an advantage.

It’s the same situation as with nuclear bombs, you cannot uninvent them. We just have to live with the fact that because of their existence, our world is permanently an uglier place.

5 Likes

IMHO, it’s not the same, which I wanted to express in my 1st post.
The invention is not new. It was always there for decades.
Baby monster needs food to grow up, which is computing power (super computers) and data (human behavior).
While super computers is not in our power to stop (billions of $/€ invested), feeding the beast with human behavior is only at free will.

I am no conspiracy theorist, nor blind fanatic Christian, but this looks horribly familiar to the Apocalypse, IMO.

1 Like

To me it seems like, at some point, we died and went to hell, but didn’t realise it. And here we are, and everything looks exactly the same.

5 Likes

It’s a tool and it’s on yours will to decide how to use it.

But at first, with all this “monster feeding”, even posting here is feeding countless AI algorithms and will continue to. It’s inevitable, and it’s won’t be stopped.

Even in best anti-AI scenario with whole world thermonuclear bombardment orsuper strong Solar storm there’s a one huge problem - organic AI are a thing and that thing is learning quicker than a computers.

So instead of inorganic AI overlords we could get horrors "beyond human comprehension"™ like in Warhammer, where even toster has it’s own brain in a Vat.

You know how it goes… :rofl:

image

3 Likes

Think twice!

Independent fact-checkers… :rofl:

3 Likes

Meh, not seems even close to be a 3 years of the antichrist reign.
There’s neither truth humans, neither in machines.

Not to say, from the Christian Orthodox point, Apocalypse is a good thing, as it’s an inevitable ending of this sinful world and ultimate defeat of evil.

Though I personally think if even half of the angels once fall, so humans will too even after the Apocalypse.
So there’s will be only war, with neverending fall into the sin.

Two interesting and unbiased / informative in-depth articles on the whole topic by Scott Alexander:
(1) Mostly Skeptical Thoughts On The Chatbot Propaganda Apocalypse
and
(2) OpenAI’s “Planning For AGI And Beyond”

1 Like

While super computers is not in our power to stop (billions of $/€ invested), feeding the beast with human behavior is only at free will.

I think in the end there will be constant warfare competition like with any other local/selfhost solution vs cloud one.

And currently, tools like Stable Diffusion and LLaMA are showing to be far more efficient while running even on a singular PC. 7B fine-tuned LLaMA(Alpaca) is somehow able to work on a damn raspberry pi4.

I’m a huge fun of talking religion, but I think it would ruin this topic. I suggest we stay on technical non-spiritual arguments, IMHO.

Not the same. Different applications, different outcome per computing power, to my knowledge.

This is one of the wished advertised projects, that are only created to confuse and frighten people. Electronic games is the greatest source of the primary data collection on human behavior. People play games and feed the AIs for years. Now the more real world data collection is required. That’s why this huge promotion of ChatsGPTs was introduced.

1 Like

Imagine if this 2 just didn’t ate a damn fruit and didn’t listen to some sneaky serpents… :smiling_face_with_tear:

If to look on the state of the things without faith, with all modern understanding of the cyclic universe/multiverse, it all becomes even more like neverending hell. And so on infinite scale of cycles even if there will be a periods like absolutely lifeless universe, all of the events have non-zero chances to be repeated at least locally on our tiny sphere of the dead stars ashes, that’s floating in the void of darkness.
And so in eternity of the strange eons, this non-zero chances becoming an infinity.

It could be a coincidence, but clicking/opening this URL broke my session (picom and/or Xorg).

They are watching!!! :scream_cat:

4 Likes

Not the same. Different applications, different outcome per computing power, to my knowledge.

In the matter of feeding they’re pretty the same. They’re just scrapering all data available, while swing on “Feed the Machine”.

To comprehend the level of the cyberpunk dystopia we’re living in, from about ~2015 smart TVs and other devices are actively using a damn ultrasound to communicate with each other to collect more data for the personalized ADs an to make sure that yours FBI agent won’t miss anything about you.
Like if they have not enough ways to create a user ID in just a web already, with all this tonns of cookies and countless javascript trackers.

This is one of the wished advertised projects, that are only created to confuse and frighten people. Electronic games is the greatest source of the primary data collection on human behavior. People play games and feed the AIs for years. Now the more real world data collection is required. That’s why this huge promotion of ChatsGPTs was introduced.

The thing is, such brains in a vat are basically still humans, though are being trained like digital AI.
Funny how neural networks made a full cycle as in the beginning they were a math models to describe organic neural systems.

Biotechnological side is also quite impressive, manipulations with living flesh after discovery of Yamanaka factors in 2000s are gone so far, that there’s even no need in embryos. They’re using reprogrammed adult cells and transforming them into the neurons.

It’s hilarious that we still barely can create proper brain-to-computer interface for a proper human being. So it’s easier to use a grown in a vat mini brains.
But I’m pretty sure in the future this mini-brains gonna turn macro and will work in giant data center like silicone machines right now.