rubystillshits

@RubyStillShits

The shittier page of @missrubyfarts

Las Vegas

Joined July 2024

25 Following

3K Followers

602 Posts

rubystillshits @RubyStillShits

about 13 hours ago

I have a new fan who I will take care of very well. Do you want it? Subscribe to my Scatbook channel at https://t.co/C9yt7iry4c

RubyStillShits retweeted

Miss Ruby Farts LAS VEGAS

@missrubyfarts

about 17 hours ago

Guess the size of my massive mommy milkers in the comments!!! If you know my band or cup size, no cheating! I want to send a topless selfie to anyone who guesses correctly.

RubyStillShits retweeted

Miss Ruby Farts LAS VEGAS

@missrubyfarts

about 17 hours ago

3 years ago I was 36DDD, yes I am making this easier on everyone so I can show off my current tits that are EVEN BIGGER!! (📸 @Nikki_Sequoia)

missrubyfarts's tweet photo. 3 years ago I was 36DDD, yes I am making this easier on everyone so I can show off my current tits that are EVEN BIGGER!!
(📸 @Nikki_Sequoia) https://t.co/61HaMMzZlL

rubystillshits @RubyStillShits

6 days ago

I have a new fan who I will take care of very well. Do you want it? Subscribe to my Scatbook channel at https://t.co/C9yt7iry4c

253

rubystillshits @RubyStillShits

9 days ago

I have a new fan who I will take care of very well. Do you want it? Subscribe to my Scatbook channel at https://t.co/C9yt7iry4c

345

RubyStillShits retweeted

Miss Ruby Farts LAS VEGAS

@missrubyfarts

15 days ago

The @GPTeaseAI situation in a nutshell: The support rep’s responses reveal a few structural problems at once: weak escalation handling, defensive framing, and a mismatch between the product’s promises and its actual behavior. All for 19.99/month!

RubyStillShits retweeted

Miss Ruby Farts LAS VEGAS

@missrubyfarts

15 days ago

Part 2: The support interaction also revealed a broader customer service problem. My original complaint was straightforward: the system fabricated traits about me that I never provided and continued doing so even after repeated corrections. Instead of directly acknowledging or investigating that issue, support repeatedly redirected the conversation toward my persona configuration, effectively reframing a system behavior problem as user error. The repeated explanation that “AI always hallucinates” felt dismissive because hallucinations in general are not the same thing as persistent failures in memory consistency, instruction hierarchy, or user identity grounding. A personalization-focused AI product should be optimized specifically to reduce those kinds of failures. The response that “persona instructions override chat instructions” unintentionally highlighted the deeper design flaw because if persona memory retrieval is unreliable, stale, or contaminated, then users lose the ability to meaningfully correct the system in-session. The conversation became increasingly frustrating because my attempts to precisely define the issue were met with generalized explanations instead of technical accountability or escalation. Even when I clarified that small inaccuracies were acceptable but fabricated explicit physical traits were not, support continued responding as though the problem was merely expectation management. The later suggestion of a walkthrough or onboarding session also felt premature because there was still no clear explanation of what additional setup would actually solve if the underlying memory system itself was unreliable. Ultimately the experience created the impression that the product’s marketed value proposition — personalization and memory continuity — was not functioning at a level that justified a paid subscription, especially when free alternatives were producing more consistent and reliable results. The support rep’s responses reveal a few structural problems at once: weak escalation handling, defensive framing, and a mismatch between the product’s promises and its actual behavior.

RubyStillShits retweeted

Miss Ruby Farts LAS VEGAS

@missrubyfarts

15 days ago

Im still waiting on my DM to be replied to, in the mean time I thought I would offer a public response to what has been going on with @GPTeaseAI in my DMS. My issue was never that the AI made occasional mistakes in general, it was that the system repeatedly invented traits about me that were never in my persona while also ignoring direct corrections and constraints, and instead of addressing that actual failure mode, support kept reframing the issue as though my persona setup itself was the problem. Saying “AI will always hallucinate” misses the point entirely because I wasn’t asking for perfect intelligence, I was asking for basic consistency in a paid personalization product that is specifically marketed around remembering users and maintaining persona continuity. There’s a major difference between a model making a harmless mistake like getting a hair color wrong and a system repeatedly fabricating explicit physical traits that were never provided while overriding live conversational corrections. What made the interaction frustrating was that instead of acknowledging possible issues with memory retrieval, prompt hierarchy, stale persona embeddings, or instruction weighting, support repeatedly defaulted to explaining why the system behaved the way it did as though that excused the behavior itself. In fact, the statement that “persona instructions override chat instructions” actually highlights the architectural problem, because if persona memory retrieval is imperfect or contaminated, then the user effectively loses control of correcting the model in real time. The conversation also became increasingly defensive, especially when support implied that “other users don’t experience this issue,” which naturally reads as shifting responsibility onto the customer instead of seriously investigating the bug report. Throughout the exchange I repeatedly clarified the distinction between acceptable minor inaccuracies and unacceptable persistent fabrication because I was trying to isolate the exact failure mode, essentially doing QA triage for them, yet the responses continued focusing on onboarding and walkthroughs before even explaining what those walkthroughs would concretely solve. That’s why the entire interaction ultimately reinforced the core issue: if a paid personalization-based AI product cannot reliably remember users, obey corrections, or maintain stable identity grounding better than free alternatives, then the value proposition of the subscription itself becomes difficult to justify.