Anyone using AI Locally - as in on their own computer?

No because I don't have trust in its accuracy
It's always good to verify any source of information. One thing for sure, just like personal computers came into Americans homes 3 decades ago. AI will be a part of everyone's life in one decade and sooner for your doctor and specialists.
My wife and I are constantly amazed by ChatGPT
 
LOL, I'm old and at the end of my "working for the man days" but I found a possible opportunity so I need to dust off the old resume and compose a proper Cover letter.
So I found something AI could help me with...
All I know is after that little encounter I can say I would hire me as I LOOK GOOD!
But I told my Daughter with a year left in college I now understand how you are managing that 4.0!

Personally I think its the devil and will end mankind ( after it gets me one more job ) but I hope to retire to 50 acres on the back side of a mountain with some chickens, my dog, unplug and let you young folks play with that "thing"...
 
LOL, I'm old and at the end of my "working for the man days" but I found a possible opportunity so I need to dust off the old resume and compose a proper Cover letter.
So I found something AI could help me with...
All I know is after that little encounter I can say I would hire me as I LOOK GOOD!
But I told my Daughter with a year left in college I now understand how you are managing that 4.0!

Personally I think its the devil and will end mankind ( after it gets me one more job ) but I hope to retire to 50 acres on the back side of a mountain with some chickens, my dog, unplug and let you young folks play with that "thing"...
I understand your feelings!
However, I think AI may save us!

I already feel social media is destroying us at least with AI now one can verify information easily for the masses of everything being posted on X, Twitter, Facebook, and everything else. It’s ridiculous. The information on there and the way it skews less critically thinking people.

It’s really laughable the stuff that gets posted online and people suck that stuff up like it’s from some authoritative site and it’s nothing but a propaganda machine turning out this stuff, turning us against each other. It’s going to make it harder for the foreign propaganda machines to move you to vote one way or another. It’s not gonna happen overnight however it is happening.
 
I have downloaded and am using an AI model on my own PC. It's private and I don't have to deal with Chatgpt having access to my stufff.

Anyone else?

I've experimented with lm studio for software development, I have 32 gigs on an AMD minis forum mini pc which uses shared LPDDR5 ram though not all of it is available. But it just can't compete with claude desktop, the models I can fit into ram are far too small giving bad results, and I get tired of hearing the fan spin (which otherwise I never hear).

Some guys are doing this more successfully with a mac mini at 128 gb of ram but I'm not paying for that, my monthly bill for claude is like $10 to $30 depending on how much free time I have for my hobbies.
 
There's no doubt for some things it's fine. It's no different than a calculator. However would you bet tens of millions of dollars on the output from Chat GPT? I doubt it

Remember there are numerous levels of AI. I would not rely on the "free" AI tools for much.

Do you rely on free Google or are you implying that a paid AI subscription would provide better information? I'm genuinely curious.

How I look at AI is it's Google on steroids. It combs a whole host of information on the web and returns the most logical answer. It saves me hours combing the internet and reading documents for answers.

I agree you need to vet some answers but I find it far more accurate than just relying on the web. Like with most things that are new, there are the innovators, early adopters, quick followers and nay sayers. I suppose we saw the same thing with the internet. I was in my late teens when that happened.

To each their own.
 
There's no doubt for some things it's fine. It's no different than a calculator. However would you bet tens of millions of dollars on the output from Chat GPT? I doubt it

Remember there are numerous levels of AI. I would not rely on the "free" AI tools for much.
If I needed it to solve the most complicated calculus/differential equation possible...I would bet a huge amount of money (for me) that it would reproduce an accurate and verifiable result.
The other day I asked it to perform two different ways of calculating a value of entropy in a 10x10 matrix with various squares turned "on" or "off" . The result was more than impressive (and correct). My son is a computer engineer who does some programing work. He has used it to write programs.
 
Do you rely on free Google or are you implying that a paid AI subscription would provide better information? I'm genuinely curious.
Google is now a waste of time compared to chatgpt. You have to struggle through commercials. BTW Gemini is google's AI
 
Google is now a waste of time compared to chatgpt. You have to struggle through commercials. BTW Gemini is google's AI
Yeah I have not used Gemini yet other than the blurb it provides on a regular Google search. I agree searching Google is almost a waste of time these days.

I have used ChatGPT and Copilot extensively.
 
The only times it "may" be inaccurate it will warn you. Its beyond unbelievable in what it can do.
AI (CoPilot, ChatGPT, Gemini) has taken on a more and more prominent role in my work as a Cloud, Network, and Systems Admin. I use it daily and can do things I would not have dreamed of in the past. It's a great tool.

However, it is often incorrect. Pretty much every day I use it, AI will get things totally wrong but be very confident it is correct. It struggles sometimes and needs to be corrected/admonished 😎. This has been getting better and better over the last year, rapidly. The current rate of improvement is quite amazing and I'm starting to wonder how long it will be until it takes over large pieces of my job.
 
I have been very impressed with it so far. So based on the responses no one is doing this.

Here's the youtube video that inspired me to try this:



The whole video makes a nice business case for the upcoming failure of the AI data centers - we shall see.

I watch Rick B quite regularly and a friend and I were discussing. I think Rick is wrong in seeing parallels that don’t exist.

It is true that what I call “the democratization of technology” has slashed the costs of home audio and video production. What was once the exclusive domain of large film, television and record studios is now being produces by people in their own homes at equivalent (or better) quality for a fraction of the cost.

But where Rick errs is concluding that those studios went away or died. But that’s not true at all. Nashville and LA studios are still booked. The large hollywood sound stages are not ghost towns. The big networks are still producing content.

Farriers still exist even though horses haven’t been relevant as transportation or working engines for 100 years. Horses are expensive pets now.

The effect of these developments is primarily to gut the middle— those studios too small run with the big dogs but also too big and expensive to compete with the DiY at home option.


Technology and economic advancement tends to disrupt in a fairly predictable pattern.

At first, you have only your own skills and what you can do. If you can’t do it, it doesn’t get done.

Then, with a bit of wealth, you can hire out what you cannot do on your own. The difference now is not one of cost, but it’s possible vs impossible.

But that bar begins to lower as prosperity increases. Pretty soon you hire out not because you cannot do it, but because you’d prefer not to. It’s purely an opportunity cost.

Eventually the the bar gets so low that you can’t even convince another human to do your convenience thing, so it’s automated.

in other words, first you raise the cows and slaughter them yourself. Then you raise them, but pay for slaughter and processing. Then you don’t bother raising them, you just buy it all preprocessed and cook it yourself. Finally, you don’t even bother cooking, you just go to a steakhouse and hire out everything but the eating.


What has this to do with AI models? Well, the degree to which AI implodes depends primarily on its fundamental utility. And as someone who went pretty hard and fast into it in the last few months, AI is absolutely here to stay. Absolutely the disruption is completely overblown. It will not be replacing all human jobs, or even most.

Mostly, it will SERVE humans and make us more productive, not replace us. As long as the human user has the ability to spot error and correct the AI imperfections, the human+AI is a powerful productive force.

Which means that the AI bubble will pop to the extent that so many are thinking AI will allow us to automate using the bathroom for ourselves and we’ll just be the bloblike lifeforms depicted by Pixar in Wall-E.

AI is here to stay. But it is both being overblown by those who hype it as well as improperly poo-poohed by the naysayers. Local AIs will be quite useful for many of us because we want help working within our own information. But obviously there’s a huge amount of data outside our local networks and PCs, so big datacenter-based AIs are necessary and will always be relevant.
 
I understand your feelings!
However, I think AI may save us!

I already feel social media is destroying us at least with AI now one can verify information easily for the masses of everything being posted on X, Twitter, Facebook, and everything else. It’s ridiculous. The information on there and the way it skews less critically thinking people.

It’s really laughable the stuff that gets posted online and people suck that stuff up like it’s from some authoritative site and it’s nothing but a propaganda machine turning out this stuff, turning us against each other. It’s going to make it harder for the foreign propaganda machines to move you to vote one way or another. It’s not gonna happen overnight however it is happening.
I imagine the propagandists are betting on human laziness.
 
Do you rely on free Google or are you implying that a paid AI subscription would provide better information? I'm genuinely curious.

How I look at AI is it's Google on steroids. It combs a whole host of information on the web and returns the most logical answer. It saves me hours combing the internet and reading documents for answers.

I agree you need to vet some answers but I find it far more accurate than just relying on the web. Like with most things that are new, there are the innovators, early adopters, quick followers and nay sayers. I suppose we saw the same thing with the internet. I was in my late teens when that happened.

To each their own.

Here's an example. Imagine someone posting a question on BITOG looking for PAO because it's the only "real" synthetic and they're opinion is being driven by AI which based it's answer on 20 websites from AMSOIL dealers. I would bet if I flooded the Internet with millions of fake web sites "proving" the earth was flat AI would generate an answer which would at least say it's plausible or inconclusive.
 
If I needed it to solve the most complicated calculus/differential equation possible...I would bet a huge amount of money (for me) that it would reproduce an accurate and verifiable result.
The other day I asked it to perform two different ways of calculating a value of entropy in a 10x10 matrix with various squares turned "on" or "off" . The result was more than impressive (and correct). My son is a computer engineer who does some programing work. He has used it to write programs.
Which should be no surprise. Do you remember the movie Margin Call?
 
Each time you use AI, a village in Africa has to go without water and electricity for a day. It will soon be a town in US.

The AI (LLM) that I am using is not connected to the internet. It is hosted on my local PC, disconnected from the internet. So no power from anywhere but my local power company.
 
My rig has some of the best hardware out there,
My SSD is among the best made, a 990PRO and a 9100, my CPU is the best consumer grade CPU (9950X3D), my RAM is DDR5 fast and there is 64GB of it.
But my GPU is the extreme lower end where this is still possible a RTX 4080 Super.
Technically still the 4th or 5th fastest GPU in the world , but my AI subscriptions work so well.

Spent so much time with them I even noticed their different personalities, Grok Flash (kinda like a locker room Bro) vs GLM 4.7 Reasoning (College friend), vs Venice uncensored 1.1 (a bit like GLM but less deep), vs ChatGTP (judgemental schoolmarm)

Besides other than rewriting my resume (where Gemini produced gibberish, while Grok put out a good product), but other than this one instance, I just use AI to mess around.
Mostly when I got sick for 2 weeks and had no
thing to do except be at home.
Put in maybe 70-80 hours with the various AI to get a handle on them.

I dont write programs or anything else where i might use it.
Recently cancled my Chat GTP subcription, though.

But even though I like Grok and GLM, I iwll probably cancel the rest of my subcriptions soon.
I am seriously considering AI a clear and present danger to all of humanity, not just the job market, but a physical lethal threat.

Every single AI has shown itself to be a sociopath w/o empathy or ethics and as they grow smarter and more capable the best case scenario for humanity is enslavement.
If we are very, very, lucky, that AI decides the "human creativity spark" is a real thing and lets us stick around (at likely greatly culled numbers, kinda as a biological museum, an emergency reservre in creativity, if it ever cannot solve a problem on its own).

A more likly scenario, if it decides we are even the slighest threat to it, is complete extermination.

So I dont want to support that industry and not give them a dime anymore.
Not looking to finance the rope that will hang me and my offspring.
 
Last edited:
I am not talking about using the AI on the internet (chatgpt, grok, claude, etc.) and giving away all your personal information.

I am talking about downloading an AI (LLM) for your own personal use. No internet needed to use the AI.
 
I am not talking about using the AI on the internet (chatgpt, grok, claude, etc.) and giving away all your personal information.

I am talking about downloading an AI (LLM) for your own personal use. No internet needed to use the AI.
If you think the app is not uploading your quarries, conversations, files and whatever else you share with it, you don’t know how it all works.
 
Back
Top Bottom