Raw LLM Responses
Inspect the exact model output for any coded comment.
Look up by comment ID
Random samples — click to inspect
G
Buddy read the latest research paper by anthropic on LLM models. You will unders…
ytr_Ugx3Ni37n…
G
I disagree. A robot can replace an assembly line position. AI is a tool used by …
ytr_UgyRbRiHS…
G
If humans-wait. WHEN human build a robot at this level. You will find the robot …
ytc_UgxEFP5Cm…
G
Bro these scientists swear they know everything but there's certain stuff you ju…
ytr_Ugw6LTEr2…
G
Lmao, ok? AI use existing materials and artstyle by several artists to generate …
ytr_UgwcpE_OO…
G
Satoshi nakamoto ist just an AI and we build the brain for it, thinking mining b…
ytc_UgxMm4nQe…
G
If I saw this I would call the cops on the robot then break it…
ytc_Ugxxvs6sh…
G
Option 4: AI economy goes busts, everyone loses their jobs, everyone has to bail…
ytc_UgzZ7OXqy…
Comment
This is outdated and inaccurate in some important ways. Hallucinations are not caused by AI not understanding what it is saying. An LLM is a token prediction mechanism. It has no capacity to "understand" anything. Hallucinations are caused by variances in the batch size (ammount of data processed at one time) the next token is predicted with and temperature settings (the probability range the next token generated will be). The major issue that is being highlighted here is an LLM's number one weakness: non-determinism. This means that it is impossible to debug any one bug and then implement that fix for similar bugs in the conventional maner. By using a fixed batch size and a 0 temperature you can create a purely deterministic LLM as shown by Thinking Machines in their blog post on September 10th of this year. This will result in the vast majority of the issues cited here being solved because it all boils down to one core issue. Previously you could not effectively debug a LLM and now you can.
youtube
AI Responsibility
2025-10-03T16:3…
Coding Result
| Dimension | Value |
|---|---|
| Responsibility | none |
| Reasoning | consequentialist |
| Policy | none |
| Emotion | indifference |
| Coded at | 2026-04-26T23:09:12.988011 |
Raw LLM Response
[
{"id":"ytc_UgynfEijUvzZe0ZqF3V4AaABAg","responsibility":"company","reasoning":"deontological","policy":"regulate","emotion":"outrage"},
{"id":"ytc_Ugw7CQLpJ1FPVqf_d_l4AaABAg","responsibility":"none","reasoning":"consequentialist","policy":"none","emotion":"indifference"},
{"id":"ytc_Ugz9jSxtu37K-mdjEZd4AaABAg","responsibility":"distributed","reasoning":"consequentialist","policy":"none","emotion":"resignation"},
{"id":"ytc_UgzI76bty-Vihfexy8N4AaABAg","responsibility":"company","reasoning":"consequentialist","policy":"liability","emotion":"outrage"},
{"id":"ytc_UgwSFCw_0ZNBCr5KYqJ4AaABAg","responsibility":"company","reasoning":"deontological","policy":"regulate","emotion":"outrage"},
{"id":"ytc_UgxCnkHlQ0JnxyYWBgF4AaABAg","responsibility":"company","reasoning":"consequentialist","policy":"industry_self","emotion":"mixed"},
{"id":"ytc_Ugx8gCQANMHqGcsVoi94AaABAg","responsibility":"none","reasoning":"consequentialist","policy":"none","emotion":"indifference"},
{"id":"ytc_UgyfF1_xlEH-8xrHFjZ4AaABAg","responsibility":"company","reasoning":"consequentialist","policy":"industry_self","emotion":"approval"},
{"id":"ytc_Ugx74C8wtpt97sedo8R4AaABAg","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"mixed"},
{"id":"ytc_UgxsOlEqqzcYytpaEDV4AaABAg","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"approval"}
]