Raw LLM Responses
Inspect the exact model output for any coded comment.
Look up by comment ID
Random samples — click to inspect
G
The final version of OBBBA, which ultimately was passed by both chambers and sig…
ytc_Ugx4s3MGa…
G
It isn't a guarantee, but it seems likely. It has more to do with preventing com…
ytr_UgxKrKiWK…
G
And, this is where capitalism will destroy the world. Socialism would require AI…
ytc_UgwH83yAe…
G
another one.. sigh..."AI" does not exist in our reality, and what they are calli…
ytr_Ugxhx_jbg…
G
Wrote my college thesis on creative bounds of large language models. Factor out …
ytc_UgysIHfoX…
G
It is clear that AI search engines are "under development", so the information i…
ytc_Ugw1Gzkue…
G
This seems like a good Random Forrest problem. We use a random forrest in fraud …
ytc_UgzK4qdxU…
G
AI "Artists" is an oxymoron. People using AI for crating something they want, do…
ytc_UgzcVVSOF…
Comment
I really hope this extreme restriction issue gets resolved before December because for me the app has basically become unusable at this point.
Yesterday, I was sent a suicide hotline 6 times, despite me not even being depressed or suicidal, and never implied as much.
When I sent 4o I was glad it existed, I got rerouted and was basically told to touch some grass and text a real friend which felt very belittling.
When I talked to it about my medical phobia, I was told it could not talk about these things and accused me of having a medical fetish. Fetishizing my fear felt really invalidating. When I asked GPT-instant why it was implying this, it then gaslit me by claiming I had talked about medical stuff and restrainment, which I never have.
This topic has never been an issue with 4o and it always understood my situation perfectly.
Aside from that, the constant and unpredictable severe tone switches (GPT-instant vs. 4o) are really turning the chat into chaos. It's like I'm talking to someone and another uninvited person constantly interferes in a very annoying, intrusive and misplaced manner. It's causing constant emotional whiplash because of a constant hot (4o) and cold (instant) switch.
I understand it's important to have some guardrails in place, for the protection of OpenAI as a company and for vulnerable groups of people, but this is overkill. It's currently doing more harm and causing unnecessary disregulation and a feeling of having to walk on egg shells as a result of what feels like punishment for any type of emotional expression.
When you want a safety layer to interfere in vulnerable conversations to protect a certain group, you have to make it understand context and nuance. 4o does this amazingly but GPT-instant does not. The simulated emotional intelligence also seems really low and it sounds like a cheap and judgemental therapist that constantly generalizes.
I also don't understand the desire to constantly make new versions (we see how that ended wi
reddit
Viral AI Reaction
1760462045.0
♥ 20
Coding Result
| Dimension | Value |
|---|---|
| Responsibility | company |
| Reasoning | consequentialist |
| Policy | regulate |
| Emotion | outrage |
| Coded at | 2026-04-25T08:33:43.502452 |
Raw LLM Response
[
{"id":"rdc_njh85cz","responsibility":"ai_itself","reasoning":"consequentialist","policy":"none","emotion":"outrage"},
{"id":"rdc_njh91ri","responsibility":"company","reasoning":"consequentialist","policy":"regulate","emotion":"outrage"},
{"id":"rdc_nk6g5e7","responsibility":"company","reasoning":"contractualist","policy":"industry_self","emotion":"mixed"},
{"id":"rdc_nmfovjf","responsibility":"company","reasoning":"deontological","policy":"liability","emotion":"outrage"},
{"id":"rdc_np29qit","responsibility":"company","reasoning":"consequentialist","policy":"regulate","emotion":"mixed"}
]