Raw LLM Responses

Inspect the exact model output for any coded comment.

Comment
I work in the FR industry... AT BEST, FR is about 90% accurate... In order to get that level of accuracy, cameras must be positioned at a correct angle (between 30 and 45 degrees) at a height of 10-12 feet. Surprisingly, image clarity is not as significant as you might think... That said, most of our customers average about a 75% accuracy. Currently, FR is based of an algorithm that assigns a number based value to a individual face and then compares that number against other numbers to find "similar" results... It cannot identify a unique mole, scar or anything of that nature, although it will take it into consideration in regards to the assigned number value. Point is, using FR for law enforcement without a human to validate the results is ludicrous... But the fact remains that facial recognition will NEVER be 100% accurate and there will always be mistakes. It's up to us as a society to rectify those. In this case, yes, he was wrongly arrested... But then human intervention determined the FR was wrong and the mistake was corrected. It's a tool that can be used for good or evil... Just like anything else.
reddit AI Harm Incident 1593030994.0 ♥ 86
Coding Result
DimensionValue
Responsibilitycompany
Reasoningunclear
Policyindustry_self
Emotionindifference
Coded at2026-04-25T08:33:43.502452
Raw LLM Response
[ {"id":"rdc_gt6xfv2","responsibility":"government","reasoning":"deontological","policy":"regulate","emotion":"approval"}, {"id":"rdc_gt7jkhs","responsibility":"none","reasoning":"mixed","policy":"none","emotion":"outrage"}, {"id":"rdc_gt5hn7i","responsibility":"user","reasoning":"consequentialist","policy":"liability","emotion":"outrage"}, {"id":"rdc_fvw0s3l","responsibility":"company","reasoning":"consequentialist","policy":"industry_self","emotion":"indifference"}, {"id":"rdc_fvwmw8o","responsibility":"company","reasoning":"deontological","policy":"regulate","emotion":"fear"} ]