Raw LLM Responses
Inspect the exact model output for any coded comment.
Look up by comment ID
Random samples — click to inspect
in
Honestly, the part that stood out to me wasn’t how much energy a single AI query…
7468632676229…
in
Václav Šulista thank you for always supporting the mission and being the first t…
7465078930551…
in
AI in research should feel like an AGI as removing the human from the loop might…
7466859369691…
in
Hey Ruben Hassid - last week Claude hallucinated data when I asked it to analyze…
7464862742315…
in
Strong explanation. Many people underestimate that enterprise AI is not just abo…
7464978030877…
in
The next phase of AI strategy and implementation will be discerning where to use…
7470166502525…
in
Kun Cheng Absolutely. Physical AI becomes truly useful only when intelligence mo…
7463744911222…
in
Shipping 8x more code is just faster interpolation, not true recursive self-impr…
7468747076127…
Comment
This gets me thinking in lot of other examples of hidden inference Name in prompt: “Help John Smith with his resume” vs “Help Priya Sharma with her resume.” AI suggestions subtly shift, different tone, different industry assumptions. Writing style: Formal academic English vs casual slang. Same question, different answers. AI adjusts confidence level, complexity, even what it omits. Currency/units: Type “$500 budget” vs “₹500 budget.” AI changes scope of recommendations entirely, not just currency conversion. Time format: “Schedule at 3pm” with no timezone. AI infers timezone from language/locale context, silently. Gender pronouns in context: Describe a nurse vs describe a surgeon. AI completion biases shift based on training data stereotypes, even when not asked. Looks like our system prompt keeps getting bigger
LinkedIn
AI Safety & Risk
Director - Data & AI (APAC)
2026-05-31T08:1…
Coding Result
| Dimension | Value |
|---|---|
| Primary value | fairness |
| Secondary value | transparency |
| Alignment target | individual_users |
| Stance | critical |
| Emotion | outrage |
| Value justification | The speaker wants AI to treat individuals equally regardless of their name, language, or cultural context, which aligns with the value of fairness. |
| Target justification | The target of the speaker's concern is individual users, such as John Smith or Priya Sharma, who may be affected by AI's hidden inferences and biases. |
| Coded at | 2026-06-11T08:33:02Z |
Raw LLM Response
```
{
"value_primary": "fairness",
"value_secondary": "transparency",
"target": "individual_users",
"stance": "critical",
"emotion": "outrage",
"value_justification": "The speaker wants AI to treat individuals equally regardless of their name, language, or cultural context, which aligns with the value of fairness.",
"target_justification": "The target of the speaker's concern is individual users, such as John Smith or Priya Sharma, who may be affected by AI's hidden inferences and biases."
}
```