Raw LLM Responses
Inspect the exact model output for any coded comment.
Look up by comment ID
Random samples — click to inspect
G
lol. They’re going to have the snarkiest, most “well ackshually” annoying, and h…
rdc_kr5dwyl
G
there needs to be a term like 'dunning-kruger effect' but for people that think …
ytr_UgzBOXooJ…
G
Yup. This is the standard new tech business model. Put out a great product at a …
rdc_mrtzc7h
G
Lets do away with AI if it's going to cause this much damage, what are we stupid…
ytc_UgxjbO_a9…
G
depends on the model of nightshade, if you read on this you will now that right …
ytr_Ugxvfv4ti…
G
We all need to be aware of internal dialogue. I liked both the interviewer and G…
ytc_UgxmshTHx…
G
It could happen if ignorant CEOs don't really know what the capability of AI is …
ytc_UgxzRMYyq…
G
Elon is a smart man wow AI we have to be careful with this AI intelligence…
ytc_UgxiEQvyo…
Comment
Saying that "nobody knows what's going on inside an LLM/AI" is plain wrong and displays a fundamental lack of knowledge.
From a purely mathematical perspective, we already know exactly what happens inside an AI system. It follows predetermined algorithms that are precisely defined by a computer program, as well as training data that is also processed by well-specified algorithms.
And these very algorithms are the trade secrets of the companies that release LLMs/AIs. One can safely assume that these companies have deliberately implemented debugging mechanisms to allow for detailed analysis of their program algorithms.
The claim that 'nobody knows what's going on inside an LLM/AI' is absurd. The companies certainly know very well. Because they can easily build in barriers that, for example, suppress antisemitic responses.
Eliezer, you are wrong. AI can be controlled the same way as nuclear weapons are. AI cannot emerge uncontrolled in some dark cellars and cannot control humans by bribing them on a large scale threatening mankind without going unnoticed. This is Sci-Fi at its best, to say mildly.
youtube
AI Governance
2025-08-08T17:3…
Coding Result
| Dimension | Value |
|---|---|
| Responsibility | developer |
| Reasoning | deontological |
| Policy | unclear |
| Emotion | outrage |
| Coded at | 2026-04-27T06:24:53.388235 |
Raw LLM Response
[
{"id":"ytc_UgzCnJEYgS4Gt6aF4mt4AaABAg","responsibility":"none","reasoning":"unclear","policy":"unclear","emotion":"indifference"},
{"id":"ytc_Ugy2SpFDrQzFraxnnKd4AaABAg","responsibility":"distributed","reasoning":"consequentialist","policy":"unclear","emotion":"fear"},
{"id":"ytc_UgyOqt8CXQJ8RdsMO9F4AaABAg","responsibility":"ai_itself","reasoning":"consequentialist","policy":"unclear","emotion":"outrage"},
{"id":"ytc_UgyOOTryi47R_beqi6Z4AaABAg","responsibility":"none","reasoning":"mixed","policy":"unclear","emotion":"mixed"},
{"id":"ytc_UgyIAa6SDkpXaA0j1dJ4AaABAg","responsibility":"developer","reasoning":"virtue","policy":"unclear","emotion":"mixed"},
{"id":"ytc_UgymNBA5-cA903RN8d94AaABAg","responsibility":"none","reasoning":"unclear","policy":"unclear","emotion":"approval"},
{"id":"ytc_UgySrkaLrZkp5tjm5ft4AaABAg","responsibility":"none","reasoning":"consequentialist","policy":"unclear","emotion":"resignation"},
{"id":"ytc_UgwfKuG-8cR_batwnEh4AaABAg","responsibility":"developer","reasoning":"deontological","policy":"unclear","emotion":"outrage"},
{"id":"ytc_UgzLC6OzO5UJWwE3sft4AaABAg","responsibility":"developer","reasoning":"deontological","policy":"unclear","emotion":"outrage"},
{"id":"ytc_UgwCFO_EfVHH2DcNgUJ4AaABAg","responsibility":"user","reasoning":"consequentialist","policy":"unclear","emotion":"resignation"}
]