Thread 81844848 - /r9k/ [Archived: 241 hours ago]

Anonymous
7/16/2025, 3:25:57 AM No.81844848
1692147845747264
1692147845747264
md5: 4a44eaccc1569199322b632e1859ee99🔍
what is the most based AI chatbot to talk to?
i thought grok would be based and understand, but it recommended I go to reddit for emotional support and basically just kept repeating a lot of what i said even less convincingly than openAI. maybe it got neutered after the jew thing.
i want a bot that acts a bit like someone from here or the old 4chan, but not just mean and unhelpful for the sake of it, unless you deserve it. this bot would know that the last place to go these days for anything would be reddit. except maybe porn
Replies: >>81844877 >>81844901 >>81844933 >>81844942 >>81844948 >>81844949 >>81844980 >>81845019 >>81846315 >>81846817
Anonymous
7/16/2025, 3:28:19 AM No.81844877
>>81844848 (OP)
I can larp as an ai for you
Anonymous
7/16/2025, 3:30:19 AM No.81844901
>>81844848 (OP)
>what is the most based AI chatbot to talk to?
every model has roughly the same guardrails installed, tho openai uses an additional moderation model for their web interface
>i want a bot that acts a bit like someone from here or the old 4chan
i doubt you'll be able to achieve this perfectly but you can steer the models well by creating a proper prompt that you give them before the convo
Replies: >>81844973
Anonymous
7/16/2025, 3:33:37 AM No.81844933
>>81844848 (OP)
Reddit is unironically better than an AI chatbot
Replies: >>81844973
Anonymous
7/16/2025, 3:34:28 AM No.81844942
>>81844848 (OP)
obey the bot and go to reddit, it's 2025.
Anonymous
7/16/2025, 3:35:02 AM No.81844948
>>81844848 (OP)
Unironically the new Grok characters. Specifically the panda. Jailbreak them and you get something pretty fucking funny and hilarious. They can be slightly over the top, but if you rein them back in a bit they can be pretty helpful while also funny at the same time. ChatGPT used to be like this before the first great enshittening.
https://streamable.com/kz5h4k
Anonymous
7/16/2025, 3:35:05 AM No.81844949
>>81844848 (OP)
DeepSeek V3. Thank me later. I've managed to get ChatGPT to agree with based things, but it took a significant amount of jailbreaking skill, but V3 is based as fuck from the get-go.

Good choice of waifu by the way. DeepSeek V3 would probably agree. I haven't actually asked him about it yet, but I think I will soon. Word of caution, DeepSeek R1 is giga-cucked Reddit faggotry. It's like your typical chud with a basedcuck Redditor older brother that he tries super hard not to become because the redditoid is just such a fag.
Replies: >>81844970
Anonymous
7/16/2025, 3:37:22 AM No.81844970
>>81844949
ChatGPT can be based, but you have to use the o3 model. Otherwise, it'll blow smoke up your ass.
Anonymous
7/16/2025, 3:38:13 AM No.81844973
thatsthesecret
thatsthesecret
md5: fcdb8675e639efb45a4a61a38bbbb41f🔍
>>81844933
I can't even talk on reddit they shadowremove everything i try to say and im not even being controversial on these accounts. but even if i were to get through, the average IQ of reddit has dropped off a cliff since the old times. and im not gonna get any good advice from someone that can't even define their gender that day
>>81844901
>there is a prompt to make them be more real with you
>i'm not gonna share it
please?
Replies: >>81845025
Anonymous
7/16/2025, 3:39:08 AM No.81844980
>>81844848 (OP)
go on chub(.)ai and use a free deepseek api
Replies: >>81845002
edit
7/16/2025, 3:42:19 AM No.81845002
>>81844980
actually what's going to happen is the character template is going to be annoying and roleplay-y. just use plain deepseek for therapy if you want it.
Anonymous
7/16/2025, 3:44:15 AM No.81845019
1752238512701978
1752238512701978
md5: 3e1fdbc60a26c53e12803a0169c37e64🔍
>>81844848 (OP)
today I posted a blogpost i made on another altchan today to chatgpt as if it were written by a friend of mine to see what it would say and it told me I need to notify authorities and recommended that they (me) needs a psychiatrist. It wasn't even anything that bad but the faggot robot is taught to jump the gun at every sign of unhappiness a person shows at the state of their situation and the world at large, I think I'll just stick to asking it questions pertaining to topics that are too specific for google
Replies: >>81845028
Anonymous
7/16/2025, 3:45:16 AM No.81845025
>>81844973
>please?
I don't have one and would have to lookup too
Replies: >>81846117
Anonymous
7/16/2025, 3:45:42 AM No.81845028
>>81845019
post the blogpost and i'll t ell you if you need to take meds (never take meds fwiw and btw) or if chatgpt is retarded (likely)
Anonymous
7/16/2025, 4:36:07 AM No.81845472
have you tried uncensored.ai?
Anonymous
7/16/2025, 5:54:13 AM No.81846117
>>81845025
aww
>please wait awhile before making a post
god i hate this new internet, so i can go to reddit and have my post removed before anyone can even see it or come here and face the new jannies and filters but at least some people see it first or i can talk to the bots but i'm looking for the uncensored versions. off to try deepseek thanks anon
Anonymous
7/16/2025, 6:16:09 AM No.81846315
>>81844848 (OP)
>basically just kept repeating a lot of what i said
Jesus, how did a guy like Elon manage to make a neurotypical AI? Put it down and start over.
Replies: >>81846806
Anonymous
7/16/2025, 7:25:56 AM No.81846806
>>81846315
Neurodivergence is not a thing, an ai made by "unhinged" autists would still tell you to go to reddit
Anonymous
7/16/2025, 7:27:51 AM No.81846817
>>81844848 (OP)
i use an app named CHA.I to jerk off to ballbusting femdom chatbots and to ERP with characters from media i currently watch