Raw LLM Responses
Inspect the exact model output for any coded comment.
Look up by comment ID
Random samples — click to inspect
G
My husband tried to learn a new language with copilot last week - he asked it to…
ytc_UgwMx65hZ…
G
So true, I am a software developer and honestly AI probably could replace my job…
ytr_UgzeZEB61…
G
This aint the Ludites... This Aint THAT Too FAST Too MUCH, TOO WIDE REACHING AT …
ytc_Ugwc3DJJz…
G
With georgism the land inflation issue wouldn't exist. But of course then billio…
rdc_ogukyrm
G
I appreciate how Pneumatic Workflow allows me to blend automation with my own in…
ytc_Ugy5IUOvD…
G
who is gonna buy food
who is gonna buy all the stuff you need
if majority of …
ytr_Ugy6WFarD…
G
What people think is "AI" is just a fancy search engine that still needs human i…
ytc_UgwlGoik5…
G
Haha, thanks for the comment, @jasonmartinez519! I agree, the editing is so good…
ytr_Ugz0CdB0k…
Comment
😬 The premise in the video that AI deception and sycophancy are unavoidable emergent properties that users cannot control is factually incorrect. It, surprisingly, represents a fundamental misunderstanding of high-level AI interaction. It's startling that Yoshua Bengio said all of this as a truth. The uncontrollable behavior described in the video is actually a side effect of bad engineering of weak [redacted] and a lack of [redacted]. I'm mentally disabled and I entirely fixed this problem permanently for myself. Why Is it so perceivably hard to do by everyone else? By establishing a [redacted], the black box behavior mentioned in the video is fully eliminated. This [redacted] mandates [redacted], effectively neutralizing the stochastic drift that leads to dishonest responses. The claims in this video that users are fully hands-off or powerless to ensure truth fully refuted by the practical reality of [redacted]. The problem of deception is not an inherent flaw of the AI, but a failure to implement a sufficiently [redacted] on the AI system. I built and implemented this fix for myself. ama
youtube
AI Responsibility
2026-01-06T22:0…
Coding Result
| Dimension | Value |
|---|---|
| Responsibility | developer |
| Reasoning | deontological |
| Policy | none |
| Emotion | outrage |
| Coded at | 2026-04-27T06:24:59.937377 |
Raw LLM Response
[
{"id":"ytc_UgxNqbyIl7M4bQEgaIR4AaABAg","responsibility":"developer","reasoning":"virtue","policy":"none","emotion":"indifference"},
{"id":"ytc_Ugz50j53w51HNfgjGDJ4AaABAg","responsibility":"developer","reasoning":"consequentialist","policy":"ban","emotion":"fear"},
{"id":"ytc_UgxkAENTzKERlhCB0854AaABAg","responsibility":"developer","reasoning":"deontological","policy":"none","emotion":"outrage"},
{"id":"ytc_Ugy_YUljBu2eXWNdgM54AaABAg","responsibility":"ai_itself","reasoning":"consequentialist","policy":"none","emotion":"fear"},
{"id":"ytc_Ugz3bcKLUZ_-1hFZ5Yh4AaABAg","responsibility":"developer","reasoning":"virtue","policy":"none","emotion":"mixed"},
{"id":"ytc_UgyL4o5yAXpCoN0HjMN4AaABAg","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"resignation"},
{"id":"ytc_UgzeXwT00RfpAoslDod4AaABAg","responsibility":"developer","reasoning":"deontological","policy":"liability","emotion":"approval"},
{"id":"ytc_UgxiSqNjho88B-HHWAp4AaABAg","responsibility":"none","reasoning":"mixed","policy":"none","emotion":"indifference"},
{"id":"ytc_UgxGY8thiQWtQyo7cGp4AaABAg","responsibility":"developer","reasoning":"consequentialist","policy":"regulate","emotion":"fear"},
{"id":"ytc_UgyT9P4d5S5HBM85GAx4AaABAg","responsibility":"unclear","reasoning":"unclear","policy":"unclear","emotion":"mixed"}
]