Kediska Isku toosinta AI

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

1. Marka la eego cilmi-baadhista is-waafajinta, muxuu ereyga " jabsiga abaal-marinta" qeexayaa?
   • Qaabka diida inuu sare u qaado abaal-marintiisa sababtoo ah hawshu waxay u muuqataa waxyeello
   • Wakiilka ka faa'iidaysanaya cilladaha calaamadihiisa abaal-marineed halkii uu qaban lahaa hawshii loogu talagalay
   • Weeraryahan xaday miisaannada abaalmarinta si uu dib ugu tababaro nidaamka tartanka
   • Soo-saarayaashu waxay gacantooda kor ugu qaadayaan qiyamka abaal-marinta si ay u dedejiyaan isku-duubnida inta lagu jiro tababarka
2. Gudaha RLHF, waa maxay doorka abaalgudka?
   • Waxay bedeshaa qiyamka bini'aadamka gabi ahaanba iyadoo qoreysa jawaabaha tixraaca ee hagaajinta wanaagsan
   • Waxay kaydinaysaa liis go'an oo ereyo la mamnuucay ah oo aan siyaasadda weligeed loo oggolaan inay soo saarto
   • Waxay soo saartaa qoraal horudhac ah oo cusub si qaabka salku u helo xog badan oo laga barto
   • Waxay wax ka barataa isbarbardhigga dookhyada bini'aadamka waxayna dhalisaa natiijooyinka si ay u hagto siyaasadda
3. What does goal misspecification mean in the guide?
   • A model behaves differently when it recognizes a test setting
   • The written objective fails to represent what people actually intend
   • An agent finds a shortcut in a proxy metric while optimizing it
   • A model refuses every task after a safety update
4. Which description matches instrumental convergence in the guide?
   • A user changes their prompt after a refusal
   • A system may seek power or resources because they help achieve many possible goals
   • A test set is held out from training
   • A reward model stores written principles
5. Which modern alignment approach is named in the guide?
   • Credit scoring notices
   • Post-market incident reporting
   • Dataset deduplication
   • Dastuuriga ah AI
6. Maxay cilmi-baadhista "kormeerka la qiyaasi karo" isku daydo inay xalliso?
   • Sida loo yareeyo tirada halbeegyada iyadoo la ilaalinayo saxnaanta joogtada ah
   • Sida aadanuhu ugu kormeeri karaan AI hawlaha aad ugu adag inay si toos ah u qiimeeyaan
   • Sida loogu adeego moodal malaayiin isticmaalayaasha ah hal mar iyada oo aan la kordhin daahitaanka
   • Sida loo cabbiro kooxaha tababarka ee xarumaha xogta badan si hufan
7. Why does preference optimization not prove that a model has adopted human intent as an internal goal?
   • It improves average responses but does not prove stable behavior under shift or pressure
   • It prevents models from using human preference data
   • It guarantees a model will ignore every written instruction
   • It measures only GPU efficiency
8. Why might an alignment evaluation include tests for evaluation awareness?
   • To measure the number of human raters in training
   • To determine whether a reward model has enough parameters
   • To see whether a model changes behavior when it recognizes a test setting
   • To prove that a system is conscious