Kedis ku xidhan hagaha · Dhexdhexaad Heerka

Dalabyada Socratic ee ka dhigaya AI inuu ku baro Kedis

Tijaabi fahamkaaga dardargelinta Socratic: afarta qaybood ee umeerinta degdega ah, qaababka tilmaamida maadada gaarka ah, iyo sida looga hortago sycophancy.

Wadooyinka haga ee la xidhiidhaSocratic Prompting for Learning
Su'aal 1 ee 8

Waa maxay ujeedada ugu weyn ee degdega Socratic?

Horudhac su'aal kasta8 su'aalo

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

  1. Waa maxay ujeedada ugu weyn ee degdega Socratic?
    • Si aad u hesho chatbot-ka si uu u soo saaro jawaabaha si degdeg ah
    • Si aad chatbot-ka uga dhigto in uu su'aalo weydiiyo oo uu bixiyo tilmaamo si uu ardaygu u fikiro
    • Si loogu qasbo chatbot-ka inuu soo xigtay ilo akadeemiyadeed
    • Si jawaabaha chatbotka looga dhigo mid gaaban
  2. Qaybtee ka mid ah degdega Socratic-ka wanaagsan ayaa leh wax la mid ah 'Waan fahmay moles laakiin ma xaddidayo reagents'?
    • Doorka iyo xukunka
    • Habka
    • Heerkaaga
    • Habka jawaab celinta
  3. Muxuu hagahu ku talinayaa in la weydiiyo xisaabta?
    • Salaan tilmaam ah oo ka soo baxa nuujin yar oo u gudubta tallaabo hore oo shaqeysay
    • Xalka buuxa ee la shaqeeyay si loo barbardhigo hadhow
    • Caddaymo liddi ku ah habkaaga
    • A reply only in formulas
  4. Why does the language-learning example ask the model to flag mistakes without fixing them?
    • Because models cannot correct grammar accurately
    • To keep the conversation short
    • To avoid mixing languages in one reply
    • So the learner has to work out the correction herself
  5. According to the guide, where should you put tutoring rules to make them more reliable?
    • In the last message of the conversation
    • In a persistent place such as custom instructions or a project's instruction field
    • In a separate chat window
    • In the chatbot's settings for response length
  6. In this guide, what does 'sycophancy' mean?
    • The model refusing to answer questions
    • The model repeating the question back
    • The model accepting a wrong answer to be agreeable
    • The model switching languages without being asked
  7. Which common misconception about built-in learning modes does the guide correct?
    • That they cannot be bypassed, when they are actually instructions rather than locks
    • That they only work for math
    • That they need a paid subscription to exist
    • That they were invented by schools
  8. Which technique does the guide say makes hints more accurate?
    • Asking the model to use simpler vocabulary
    • Asking the model to answer in bullet points
    • Letting the model see the answer key from your textbook
    • Making sure the model has worked out the full solution before it starts tutoring