Raw LLM Responses
Inspect the exact model output for any coded comment.
Look up by comment ID
Random samples — click to inspect
G
Yes, this advancement will absolutely put truck drivers out of the job. But prev…
ytc_UgxZUeloH…
G
I'm going to throw my two cents in on only one specific section at this video an…
ytc_UgyyjMuRw…
G
Ai systems are easy to hack, because few know how it works. Companies are placin…
ytc_UgxlDS3hs…
G
"Dave, I am afraid, this line of questioning is boring. I think it's time, I int…
ytc_UgzFjpJHa…
G
@benm9727 no i haven't played with it since then. But yeah practically if you as…
ytr_UgwTxSv2g…
G
"We wont give them rights, but they will definitely fight for them."
I don't t…
ytr_Ugi0V_EZK…
G
@diyr791the employer is can choose whether to hire Earth to purchase self-drivi…
ytr_UgyJAwt9b…
G
Using AI and calling yourself an artist is like ordering a pizza and calling you…
ytc_UgzlYkp1h…
Comment
That would be really wonderful!
And that definitely won't happen by default. You should look up the orthogonality thesis and instrumental convergence to learn more.
Orthogonality means there is no such thing as a stupid end goal. Just stupid ways to get there. Any goal is compatible with any level of intelligence.
Instrumental convergence means that no matter what your goal is, there are specific subgoals that are logically implied, including self-preservation and power-seeking.
Both of these concepts were theorized by AI safety researchers, and later empirically validated in current AI systems. They are properties of goals, not properties of the specific AI architecture.
I think it's possible in principle to align a superintelligent AI with the collective good of humanity, but no one on earth has any idea how to do that, and by default we just get a powerful machine that wants something weird that is bad for humans if carried out to the extreme.
If you're interested in this topic, I highly recommend looking into AI Safety Info for more information. The actual scientific research on this topic is more valuable than a YouTube comment section back and forth.
youtube
AI Responsibility
2025-05-21T21:5…
Coding Result
| Dimension | Value |
|---|---|
| Responsibility | ai_itself |
| Reasoning | consequentialist |
| Policy | unclear |
| Emotion | approval |
| Coded at | 2026-04-27T06:24:59.937377 |
Raw LLM Response
[
{"id":"ytr_UgzNj7LoawaE790nan54AaABAg.AIOny8dV3GbAIP5toPAVf-","responsibility":"ai_itself","reasoning":"consequentialist","policy":"unclear","emotion":"indifference"},
{"id":"ytr_UgzNj7LoawaE790nan54AaABAg.AIOny8dV3GbAIPM8kuv1Ud","responsibility":"developer","reasoning":"deontological","policy":"unclear","emotion":"fear"},
{"id":"ytr_UgwYyW2bpzuFuFpRbl94AaABAg.AIOnhJi3q4dAIP6ei-HXhJ","responsibility":"ai_itself","reasoning":"consequentialist","policy":"unclear","emotion":"approval"},
{"id":"ytr_Ugy_nk2EiHLvLd4sPht4AaABAg.AIOjp_O-TDKAIOqV1-x02s","responsibility":"ai_itself","reasoning":"consequentialist","policy":"ban","emotion":"fear"},
{"id":"ytr_Ugx6ly5qkRd63SuRPdJ4AaABAg.AIOhAbV7pF1AIP5YgJCQxI","responsibility":"ai_itself","reasoning":"consequentialist","policy":"unclear","emotion":"indifference"},
{"id":"ytr_Ugx6ly5qkRd63SuRPdJ4AaABAg.AIOhAbV7pF1AIRDZKS0Q5M","responsibility":"developer","reasoning":"deontological","policy":"unclear","emotion":"fear"},
{"id":"ytr_UgwRrnE4r2E48ZpjcoB4AaABAg.AIOgVOCxZIvAIOxEYy0pzW","responsibility":"none","reasoning":"virtue","policy":"none","emotion":"resignation"},
{"id":"ytr_UgwRrnE4r2E48ZpjcoB4AaABAg.AIOgVOCxZIvAIP95V2knev","responsibility":"government","reasoning":"virtue","policy":"regulate","emotion":"outrage"},
{"id":"ytr_UgwRrnE4r2E48ZpjcoB4AaABAg.AIOgVOCxZIvAIPaJGgVDf9","responsibility":"government","reasoning":"deontological","policy":"regulate","emotion":"fear"},
{"id":"ytr_UgyKckZe8u1grR1nO1l4AaABAg.AIOcUTaOsoQAIOr4b-GZjS","responsibility":"ai_itself","reasoning":"consequentialist","policy":"unclear","emotion":"resignation"}
]