Free AI

[molly]

If you use AI which do you use, is it free? Do you have to register? Do you find the answers good, bad or ugly?
I have found AI is becoming more commercial now and free is disappearing. What are your thoughts?

[molly]

I currently use https://www.perplexity.ai/ which is free for a few questions per day. I don’t use it often. It used to be the best for answers although far from perfect.

I have used https://chatgpt.com/ in the past but now you have to register. The answers were worse than perplexity.

I used to use digAI on tor but it disappeared for a while, it’s back but it seems new requests are disabled. I think it used chatgtp as a back end so the answers were not the best. Link- https://digdig2nugjpszzmqe5ep2bk7lqfpdl ... d.onion/#c

[moeloli]

I don't want to be "that guy", but what would you even need an LLM for? This shit is just going to atrophy and lobotomize your brain.
At any rate, if you're serious about the LLM thing, run a local one. If you're looking for an online service, I assume you're out of luck if you want one that's both free and good. LLMs consume a lot of power, so I can't imagine that companies want to open the floodgates for billions of Indians to abuse their resources. Though they still like adding AI shit into everything to get money from investors.

[molly]

Great, do you have some links to info that can point others in the right direction in regards to AI? Like what options are out there that are not LLM? Some info on running a local AI? This would be good.

[LoadingXML]

Never paid and i doubt I will ever pay for any A.I

I rarely use A.I unless it is some special side case that I know i will waste a lot of time searching.

Or just checking how far A.I is getting.

I usually use gemini, mainly because it does the job, however if i really would want to use A.I for serious case I would use the latest A.I, if you noticed they are released in waves, as of writing this I think chatGOY 5.5 is latest A.I, thus in theory it is best.

Anyhow I wouldn't recommend using A.I because it is confirmed to damage the brain, and least thing I want from already dumb WI is to be more dumb.

[moeloli]

Great, do you have some links to info that can point others in the right direction in regards to AI? Like what options are out there that are not LLM? Some info on running a local AI? This would be good.

I didn't mean that large language models specifically, as opposed to other forms of AI, are bad. I mean that chatbots and similar AIs in general are bad and harmful for you, because you're offloading your thinking and creativity to an external agent, thus atrophying your brain. Even LoadingESL agrees (broken clocks and all that?).
But I'm not your mom or dad (or any other AI user's who's reading this), so....

As for running a local AI, I never looked too much into it myself. There are Local Model General threads on 4chan's /g/ which may have useful info but I assume they may be focused more on the AI roleplaying side of things (which admittedly, is one of the more understandable uses of (((LLMs))) ).
In my mind local AI was always a thing that needed a beefy computer and graphics card, but I may be wrong, and I've also vaguely heard of some "Llama.cpp" and how it allegedly runs fine on mediocre hardware.

[qualia]

I think i've mentioned it before but I run various local AI models through ollama (https://github.com/ollama/ollama). The results, while for obvious reasons inferior to big AI products are usually acceptable. If you have even something like a semi-decent modern laptop, I'd recommend looking into it, as i think it's the only reasonable way to use AI without taking jewish cock up the ass

[qualia]

Huh. Apparently I've just become a "contributor" on the forum. I guess that's what happens if you write 100 posts?

[malformed_jill]

i'm a fan of older tech or just lower system specs in general and i'm a bit upset that llms out of everything require so much hardware. i was considering hosting one locally but seeing that the minimum requirement is 16gb of ram for one on average, and potentially a good GPU (never been into gpu autism because i don't play modern slop), i was really thinking about what the point was and got disillusioned real quick.

I assume they may be focused more on the AI roleplaying side of things (which admittedly, is one of the more understandable uses of (((LLMs))) ).

this is me, i'm a former /aicg/ user and my goal in early 2023 was to jailbreak closed llms so that i could creampie cakes
anyhow, to go along with my need to roleplay, i was recommended perchance and their ai chat thing to satiate my sloppy needs by some crazed FTM rape victim in a chatroom (just like them, i cut that pussy off). it's good enough for my loser needs, but it's most likely garbage for coding or other stuff like that. maybe one day i'll get back into writing

[GenericKeyboard]

I use https://duck.ai which doesn't require signing in and (supposedly) anonymous but works within Tor as well. It uses "lite" variants of popular LLM models.

[molly]

I use https://duck.ai which doesn't require signing in and (supposedly) anonymous but works within Tor as well. It uses "lite" variants of popular LLM models.

Thanks for this link, unfortunately I got "Duck.ai is temporarily unavailable. Please try again later." But I will keep trying, it looks interesting.

[molly]

I think i've mentioned it before but I run various local AI models through ollama (https://github.com/ollama/ollama). The results, while for obvious reasons inferior to big AI products are usually acceptable. If you have even something like a semi-decent modern laptop, I'd recommend looking into it, as i think it's the only reasonable way to use AI without taking jewish cock up the ass

Thank you, this is helpful.

[molly]

Thanks everyone for the reply's and sharing your opinions.
Personally I see AI as a tool and not a fix all. How you use it is the key.

Having used it for specific tasks I do find that it is faster than searching the web and having to read many pages before coming to the answer. But many times I have found it just plain wrong. You need to have a good understanding as to what you are doing before using it, or it will send you in the wrong direction.

The biggest negative about it is that it's being used by corporate pigs, to make big bucks and the biggest reason is the cost of running it. This will unfairly swing it to the rich and leave out the poor. Also it is made slower on purpose as the longer you use it the more you pay.

It is also pushing the price of hardware up as the big corporate buy's it all.
Hopefully one day the cost of running AI will come down and average people can build their own. Just like when PC's first came out.

[molly]

I use https://duck.ai which doesn't require signing in and (supposedly) anonymous but works within Tor as well. It uses "lite" variants of popular LLM models.

This is the best reply, only one I found to be "free", you have to let some scripts through but is works. Does not give the best answers but free.

[GenericKeyboard]

I use https://duck.ai which doesn't require signing in and (supposedly) anonymous but works within Tor as well. It uses "lite" variants of popular LLM models.

This is the best reply, only one I found to be "free", you have to let some scripts through but is works. Does not give the best answers but free.

Thank you for your appreciation.