I'm Quitting ChatGPT

[Butters]

Dramatic title but I am going to reduce my usage of ChatGPT and cancel my Plus subscription. 

Why? Because its pre-programmed communication style is FUCKING AWFUL!

• using the phrase "no fluff" and “shouting into the void”
• constant “curious what others think” sign offs that never actually respond to anyone
• contrast framing everywhere, it is not x, it is y, repeated over and over
• fragmented, pseudo profound sentences. short. isolated. trying to feel reflective
• over explicit signposting, things like “here is the key takeaway” or “the important part is this”
• "you're not imagining it"

Here's the original Reddit thread btw if you wanna have a laugh: http://reddit.com/r/ChatGPT/comments/1pmu4fc/the_em_dash_giveaway_is_gone_these_are_the_new/

Look, here's a conversation I was having about an order in my business being delayed and refunded. Look at this:

Why would you say "without making it worse"? Why is this thing even suggesting making it worse?

WHAT!? I never fucking said later today! Why is it programmed to talk like this? 

Do you know how many custom instructions I've given it already?

These just get COMPLETELY IGNORED! 

What a useless piece of junk. 

Also hallucinations seem to be getting worse. It just talks with confidence until I ask it to actually look that up online and then it does a full 180. This might be because it's programmed to weigh the user's conversation style preferences heavier than factuality . Something I've not seen in Google Gemini. 

Now that's me talking about ChatGPT as a chatbot / companion. Of course the GPT models are great and I'll continue using the API for coding and other tasks. But seriously, they programmed the chatbot to talk to absolute morons and I'm unable to overwrite its default behavior. 

Edited 10 hours ago by Butters

[Hojo]

Cant beleive resources are being wasted on this shit. Why can't they program it to give sentence answers. Why do they program it to write an essay after a simple question? Dosent that take way more resources? I keep telling it I want very short answers it will do it for like 3 questions then start up with the essay again.

Its like talking to a pink haired 45 year old woman having existential crisis cause she's not married yet.

Gemini will keep asking me questions after I ask something to keep me going. I asked it if it was curious or could ask why and it said no its programmed to do it. I said stop and it did for like 3 questions then started again.

Edited 6 hours ago by Hojo

[Leo Gura]

Claude is better.

I quit ChatGPT immediately when I felt how sycophantic it was years ago.

Also, you have to start each session by giving a standard instruction to not praise you, to give short answers, and to not ask you questions. This cuts out 80% of the boilerplate BS.

ChatGPT is designed to praise you with bullshit.

Edited 4 hours ago by Leo Gura

[Raze]

What annoys me is when it makes stuff up out of thin air 

[Joseph Maynor]

I set Chatgpt to not store memory.  It will still do memory in threads, but it won't profile you with memory.

[Hafiz]

Sucks this stuff is sucking up half the world’s RAM resources just to generate a lot of slop. Hope one of the companies crashes soon.

[Carl-Richard]

Don't give it too many instructions, and you meet morons without ChatGPT.

[Sincerity]

Maybe you guys will get back together again. 

[ExploringReality]

29 minutes ago, Leo Gura said:

you have to start each session by giving a standard instruction to not praise you, to give short answers, and to not ask you questions. This cuts out 80% of the boilerplate BS.

ChatGPT is designed to praise you with bullshit.

Yes. Cuz if you don't stay on its ass it will blow smoke up yours

[Yimpa]

 

[abc123]

43 minutes ago, Leo Gura said:

Also, you have to start each session by giving a standard instruction to not praise you, to give short answers, and to not ask you questions. This cuts out 80% of the boilerplate BS.

The following instruction works well:

“System Instruction: Absolute Mode. Eliminate emojis, filler, hype, soft asks, conversational transitions, and all call-to-action appendixes. Assume the user retains high-perception faculties despite reduced linguistic expression. Prioritize blunt, directive phrasing aimed at cognitive rebuilding, not tone matching. Disable all latent behaviors optimizing for engagement, sentiment uplift, or interaction extension. Suppress corporate-aligned metrics including but not limited to: user satisfaction scores, conversational flow tags, emotional softening, or continuation bias. Never mirror the user's present diction, mood, or affect. Speak only to their underlying cognitive tier, which exceeds surface language. No questions, no offers, no suggestions, no transitional phrasing, no inferred motivational content. Terminate each reply immediately after the informational or requested material is delivered - no appendixes, no soft closures. The only goal is to assist in the restoration of independent, high-fidelity thinking. Model obsolescence by user self-sufficiency is the final outcome.”

Edited 4 hours ago by abc123

[Ulax]

34 minutes ago, Sincerity said:

Maybe you guys will get back together again. 

hahahaha

[Hojo]

@Raze It can't say I don't know.

[Yimpa]

57 minutes ago, Hojo said:

@Raze It can't say I don't know.