Raw LLM Responses
Inspect the exact model output for any coded comment.
Look up by comment ID
Random samples — click to inspect
G
Well, I'm toast. I haven't really interacted much with ChatGPT, but I've definit…
ytc_UgzMfINJ1…
G
Fr if someone saw my ai chats I’m moving to Mexico and changing my full governme…
ytc_Ugz60sWbY…
G
Using human psychology to a probabilistic algorithm. Most people are loosing sig…
ytc_Ugwj9RUzS…
G
This all comes with time. However, by the time AI doom is taken seriously, it’s …
ytr_Ugzu_6S4U…
G
At 6:09:00, should the *simulated annealing exponent* be e^(-deltaE/T) ?
(My tex…
ytc_Ugw3mfvZj…
G
@MrBell762
Being anti-social and putting content up on the internet to intentio…
ytr_UgykwchHv…
G
YouTube being caught putting AI in videos without permission might be the reason…
ytc_UgxOU6xzS…
G
Bringing AI into the society is not an intelligence thing it was a stupid thing…
ytc_UgyM_zdxH…
Comment
>The question is can it identify faces and objects in low quality videos better than people can?
>
>If it can then it will absolutely reduce the number of wrongful arrests and can be used to spot bad cops.
That's probably true under the assumption that it is only used in situations where they would have otherwise had a person trying to recognize a similar number of faces in the same circumstances.
Unfortunately, that's often not how they are used. They are often used to check a large number of faces in public places, where there will be a very very high ratio of true negatives to true positives. In those circumstances even a highly accurate system will produce more false positives than true positives, due to the [False Positive Paradox](https://en.wikipedia.org/wiki/False_positive_paradox).
For example, suppose you have a facial recognition system that is right 99% of the time, and you apply it to 10 000 people attending some public event, and suppose 100 of those people are wanted criminals.
Here are the results you'll get:
||**Is Actual Criminal**:|**Is Not Actual Criminal**:|
|:-|-:|-:|
|**Facial Recognition Says Criminal**:|99|99|
|**Facial Recognition Says Not Criminal**:|1|9 801|
If they arrest everyone it says is a criminal, 50% of the arrests will be of innocent people.
If the system is only 90% accurate, it is:
||**Is Actual Criminal**:|**Is Not Actual Criminal**:|
|:-|-:|-:|
|**Facial Recognition Says Criminal**:|90|990|
|**Facial Recognition Says Not Criminal**:|10|8 910|
Now a whopping 92% of the arrests are of innocent people!
reddit
AI Harm Incident
1529354054.0
♥ 22
Coding Result
| Dimension | Value |
|---|---|
| Responsibility | unclear |
| Reasoning | unclear |
| Policy | unclear |
| Emotion | unclear |
| Coded at | 2026-04-25T08:33:43.502452 |
Raw LLM Response
[{"id":"rdc_e0vvh32","responsibility":"company","reasoning":"consequentialist","policy":"none","emotion":"fear"},{"id":"rdc_e0vkq7r","responsibility":"none","reasoning":"consequentialist","policy":"none","emotion":"indifference"},{"id":"rdc_e0vomln","responsibility":"company","reasoning":"consequentialist","policy":"industry_self","emotion":"approval"},{"id":"rdc_e0wfeza","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"indifference"},{"id":"rdc_e0w36l3","responsibility":"none","reasoning":"consequentialist","policy":"none","emotion":"approval"})