The developer says some prompts use familiarity data, while others use LLM assistance followed by manual review. This describes preparation and scoring methods; it does not establish that a live chatbot judges every typed answer.
Where the answers come from
The developer FAQ describes collecting reference datasets and combining them into a cleaned answer set. Its examples include dictionaries, encyclopedias and wiki material. The question determines which references are relevant.
What AI does and does not establish
The FAQ distinguishes prompts that can use familiarity data from prompts needing model assistance and a manual pass. It does not publish a complete per-question audit, model name or live grading implementation. Treat claims about those unseen details as unverified.
Why a real answer may be rejected
A curated collection can miss a valid entry. Equally, a guess can fail the question’s restriction. Keep the full prompt, original spelling and visible response together before reporting an omission. Our rejected-answer checklist separates these cases.
Read the accepted name
Aliases can make autocorrection return a surprising entry. Check that result before interpreting its rarity. The featured Krillion pick also considers enjoyment, so it is not a promise to select the least familiar conceivable answer. See daily scoring for the separate depth rule.
Did these steps help?
Optional, anonymous analytics. Manage your choice · Add a correction
Your next steps
0/3Sources & verification
What was checked
The developer describes familiarity data for some prompts and LLM assistance with manual review for others.
Official evidence corroborated on the stated point; untested details are explicitly excluded.
Current official controls or developer explanation, with the listed secondary source. No purchase, full question-bank audit or score optimization experiment was performed.
- Krillion — official mode and controls ↗
Read-only browser inspection October 29, 2026. Access was not purchased.
- Nisu — independently inspected daily and Unlimited controls ↗
Secondary inspection of official daily rules and Unlimited access. It does not independently audit the answer bank.