I’ve been digging into the rabbit hole for a few months. Been switching to Linux, FOSS everything I can, trying to go to smaller sites, the least dubious social media, VPN, trusted mail, etc etc
But lately I’ve been in a rough state mentally (I say lately but it’s always been with me) and having GPT guide me and being able to just dictate what I think helps me a lot on various levels, for various reasons.
But at this point, using ChatGPT with a mic, isn’t this basically cancelling every effort I’ve made? (using it in the first place anyway)
I’m weak and it helps me, should I just throw my efforts out of the window and just say I don’t care about privacy anymore and use whatever everyone uses? (on one hand, I’ve found alternatives for almost everything, so I could keep on using those, but also, again, if I keep on using GPT on a regular basis, this is probably the worst threat the future has to offer in termes of privacy, so…I’m lost)
Privacy issues aside, don’t allow yourself to become too comfortable with leaning on AI for so much. Aside from the obvious things like AI info being flat out wrong sometimes and hallucinations, it’s going to train you into some bad habit holes during a vulnerable time. Look at how quickly we reach to map software for travel. It causes us to get mentally lazy.
If you are focused on using it and worried about privacy, you can host your own model like someone else mentioned, but you need a pretty beefy computer for it, and you could potentially host a model on the cloud (I know that breaks privacy and self-hosting rules a bit), but that can get expensive.
Edit: You could give Jan a try.
I’m a programmer and I’ve had to discipline myself with how I use Copilot. I try to lean on it for troubleshooting code I’ve written, and for doing tedious tasks that I know how to do but want to save time on.
I am definitely leaning on AI quite a bit lately. I ask and check sources now and then tho.
Side thinking, but I’m ot sure how we’ll see this problem in the future, when/if AI is “guaranteed” to basically never get things wrong. It’s the same with GPS/Maps software, since you’re talking about it. I’ve always been way worse than average to situate myself in space (? sorry, bad english probably), and GPS helps me immensely, me and my anxiety (and even then, I still manage to get stressed and miss things sometimes lol), it’s basically right most of the time as far as I know. Now IA being “always right” (let’s simplify) would have other consequences entirely, since it touches way more aspects of what we do as humans.
Not sure what to think from here, but yes, I’m definitely checking sources for important things, and restraining myself to not type every request I’m looking for into a chatbot.
Oh I am a terrible navigator haha. I lean on tools a ton, so it’s not faulting you. I just worry about AI because of how complex and enticing it is.
I find that if you diversify the models you use, you can find what works best for what you’re asking. I use Copilot daily at work and I have to coach it sometimes. I’ll go off on the web to check its work and correct it as needed. Sometimes it works, other times it gets into a loop of unhelpful answers.
You don’t actually have to have a beefy computer to self host. I know with deepseek there’s a 1.5b model which is optimized for low end hardware. I self hosted it on a desktop from 2012 and it worked alright. 8gb RAM, i5 processer, quad core. Was the GPU usage high? Yep. Still worked tho.
Obligatory: you’re not running deepseek 1.5B. You’re running some other distilled model which was finetuned by the deepseek folks
Wait, what? How does that work?
Deepseek 1.5B doesn’t exist. I don’t know why the Deepseek team named the models on Huggingface like this, but what is labelled as “Deepseek 1.5B” is actually not the OG Deepseek 70B model distilled to 1.5B, it’s a different model either trained or finetuned by the Deepseek team. My theory is some sort of intentional manipulation on their part so people stay confused on whether they are actually running the Deepseek model or not. There is a lot of commentary on this online, sorry I don’t have the links from the top of my head.
Oh wow, that’s awesome. I’m gonna have to try that model out. I have an old XFX 7950 I can try burning out lmao.
Use Andisearch, there are zero privacy concerns and it’s apart the most accurate (independent Benchmark)
I honestly am not sure on the privacy of it but if you really need a useful tool that’s ai driven I tend to trust perplexity.ai for general questions and research but I agree with one of the others just run ollama if you can
I get it, I’ve been there, but there is just no point in hating yourself because random big company wants to make money off of you, so they shove convenience in your face. Just do you best to align yourself to your values. Doesn’t have to be perfect. It’s ok if it’s always work in progress. Don’t put your life and wellbeing on hold.
Also duckduckgo.com/aichat is a thing.
I don’t think the AI is the problem here. I think you should see a real professional therapist.
I did, but that’s not really the subject. I’m not taking GPT as a therapist by the way, I may have been misunderstood because english is not my first language. But I’m not sharing mental health things with it.
In any case, I think the problem I’m feeling can be real for many people that are not in mental state anyway. But I totally see why you would say that, thanks.
I’m guessing GPT is for practicing your English skills because I also did this for my Spanish. Of course, that was back when I still used Google and Apple services. I’m going to try ollama, myself, but this use case is pretty specific so I wouldn’t expect an easy switch
Word, Riggs. 😆
On top of a therapist, a group session that works on Cognitive Behavioral Therapy principles or Dialectical Behavioral Therapy principles also helps because it puts you in a situation around other real people who are dealing with similar issues as yourself.
The problem (in the USA at least) is affording access to either of those things might be more complicated.
This is the actual best suggestion in the thread, though. Self-hosting an LLM is nice and all but what this poor soul really needs is therapy with a human.
It is possible to use GPT privately. Look up nano GPT. For example, that’s one way, and there are others. I want to say Eric Voorhees has a way of using GPT privately as well, but I cannot remember the name of his thing right off hand.
Duckduckgo has a way of using AI privately, supposedly. But I’m not sure how much can trust them. Thanks for the suggestion I’ll look into it
host your own AI using: https://ollama.com/
this paired with “AnythingLLM” is pretty powerful, and you dont have to worry about data being sent anywhere
I actually got it to run reasonably on a mini PC in a docker container, but my setup also lets me use CPU power as a psuedo GPU.
Llamacpp, Koboldcpp, and TabbyAPI are also popular local backends for local AI. SillyTavern or RisuAi are good frontends for a chat/RP style experience. Or LM Studio for a simple, all in one solution.
I have been suggested to self host in the past. But as I’m concerned for ecology, and let’s be a honest, a bit of my money too, I’ve thought that self hosting, meaning having the PC always on, was not a good thing. So I left the dilemma in this state, not knowing how “bad” this would actually be in terms of electric consumption, and if it was worth the trade with privacy or not.
That’s pretty dependant on hardware. If you host small stuff , like a pihole or something, that can be done relatively cheaply, by using a micro-PC or a Raspberry Pi. Some services don’t need to be always-on either, you only need stuff all the time if it’s mission-critical, otherwise you can turn stuff on and off as necessary, for power-saving purposes. Self-hosting doesn’t necessitate a huge rack and switches, or even your big gaming rig, my favorite thing to do with old laptops is throw Debian on it and find something I’d like to self-hosting from this list.
I understand where you’re coming from. I even went as far as to not using Bluetooth headphones anymore because of the radiation.I switched all my plastic utensils for wood and steel.I swapped out all my apps for foss apps.I was never really one to ever use AI for anything.I’ve kind of always enjoyed the hunt and the learning process when it came to figuring out foss applications. But it does feel super paranoid to always be worried about people watching you. When in reality, they aren’t watching you. They are using your data to make money. They don’t really care about who you are. So in short, I have started using Bluetooth headphones again, LOL.
Sounds like you might be sacrificing your mental wellbeing for some privacy gains. You should probably consider your priorities. Which one comes first?
If you say it like that…I should certainly try and remember the priorities in some situations
You see, the point is that privacy is really nice to have. However, it’s not worth it when the price you pay is your mental balance and serenity. There are easier and cheaper bits of privacy you can pursue, but you need to know where to draw the line. Some things just come with an unacceptable price. Maybe GPT is like that in your case.
You could check out localllama on lemmy to run foss ai models locally, or you could check out duck.ai as someone else mentioned. Your mental health should come first so do what you can for privacy but don’t feel bad about making compromises.
Yes, you can use an locally hosted LLM, but these never can be so powerfull as an AI with a lot of servers supporting. We have to difference it, between mere Chatbots, relying on its own database, often outdated and with the tendece to invent answers, and AI supported search engines, like Andi, You, DDG AI and others, which retrieve informations in realtime from the Web. But there is important that these are capable to difference trustworth sources from those which are not. Andi does a very good job here, because compare informations from several sites for the answer.
Ollama can pull info from the web using multiple sites, but yes local AIs are more prone to hallucination. Google did release Gemma3 which has a 27B model which is probably the most cost effective way to get into local models that rival chatgpt (if you can call about 2k cost effective). That was why I recommended duck.ai as well, as it has access to gpt and llama3.3:70b which will do a lot better.
The problem is privacythis don’t change if the AI uses sources from big bompany - Google GPTs. I tested several free AIs from Hugging face, Futurepedia and Futuretools, and since over two years I use Andi as main search, I didn’t find anything better. It use an own LLM, because Andi was the first AI search ever, former LazyWeb, long before the others released AI searches.
If I may offer a suggestion:
OpenwebUI in a docker image, cloudflare tunnel set up to enable access outside the home, the cheapest domain name you can get.
You can selfhost your ChatGPT bot so you can keep your data to yourself!
You’ll need a decently powerful PC at home, nothing crazy, any decent gaming rig from the last 5 years should run it fine!
Privacy is not all or nothing, and using one thing doesn’t cancel all efforts somewhere else.
Don’t go so far into the privacy hole that it affects your mental health. Just do what’s easy for you.
In the end you’re right, but as I was answering in another post, AI seems so big for the future and potentially invasive, that if I start giving this up, not having a google account or whatever, to me, would seem like nothing in comparison, if I’m trying to maintain a relative degree of privacy. So I feel like it’s not giving up a bit of my infos, but probably all of it, specially when using an app. Not sure if I’m making sense. But it’s like “ok, I won’t give this man my info”. “This other man, on the other hand, can live in my house and listen to everything”
You could always try out an AI service that doesn’t collect data like duck.ai: https://duckduckgo.com/duckduckgo-help-pages/duckai/
Go set up an ollama instance and melt your worries away. Data in your hands once again.
However, you should also learn some restraint. Privacy and security aren’t an all or nothing battle. Everyone has some holes in their armor, they need them to breathe.
most security planning workshops start with this statement:
“it is unreasonable as well as impossible to try and protect against every threat out there. there is no magic app, website, service that will solve everything. security is a mindset, and taking the time to understand what you want to protect and why will serve you better than any application.”
that’s a nice way to view it
Spend $500, buy a beefy used GPU and run Whisper + Mistral small. It’ll be a bit slow but now you can maintain your privacy on your own hardware! Best of both worlds.
Look for GPUs with more memory. My choice is the AMD 7900 XTX with 24Gb.
Scalpers made it expensive though
Yeah unfortunately ROCM is shit last I heard
Why?
The obsession with privacy can be taken to an extreme where it’s a negative for your mental health. You can sacrifice a little privacy for another important benefit. 100% privacy is either not possible or practical, so don’t let it stress you out. I’d be more concerned that AI might lead you down a wrong path, but you’re an adult, use your own judgement.
I get that, but the problem with this kind of AI is, I feel like I’m sacrificing a lot, not a little.
Don’t make a bigger deal out of it than it is. Nothing bad is going to happen.
What’s wrong with https://lmstudio.ai/ or https://jan.ai/ for offline, private, FOSS use?
This. It’s all incremental to me. No I can’t be 100% free of big tech and surveillance, etc etc - but how much of my data has Google not gotten since I started paying for private email 5 years ago or so?
(Not much cause everyone else refuses to leave Gmail behind 😂 but all I can do is keep moving myself in that direction.)
I feel like the “getting into privacy” journey for a lot of people tends to look like a bell curve - you start off with a few apps and minor tweaks to protect you from the worst online privacy invasions, and then it gradually builds and builds until you become the sort of person that has all their cat pictures on an air-gapped encrypted server hidden in a cupboard somewhere while you use SearX to find the best mask that will confuse facial recognition cameras, and then after a while you break through and just go back to using a few apps and tweaks to protect from the worst of it again.
100% is not possible. Everything else is an illusion.
