Kedis ku xidhan hagaha · Adag Heerka

AI ee Kediska La-taliyayaasha

Wuxuu tijaabiyaa helitaanka xudduudaha go'an, sida soo-celinta-soo-celinta dib-u-soo-celinta ay u fashilmi karto, iyo sharciyada sirta ah ee gelinta xogta macmiilka qalabka AI.

Wadooyinka haga ee la xidhiidhaAI ee La-taliyayaasha
Su'aal 1 ee 8

Maxay ku qeexeen Dugsiga Ganacsiga Harvard ee 2023 iyo tijaabada goobta BCG?

Horudhac su'aal kasta8 su'aalo

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

  1. Maxay ku qeexeen Dugsiga Ganacsiga Harvard ee 2023 iyo tijaabada goobta BCG?
    • AI made consultants worse at every task
    • A jagged frontier where AI helped on tasks within its capability but hurt on a task outside it
    • AI had no measurable effect on consultant work
    • Only senior consultants benefited from AI
  2. Waa maxay sababta hagaha uu uga digayo in isku-dhafka cilmi-baarista AI uusan ahayn mid dhexdhexaad ah?
    • Models tend to repeat the most common view and may drop outliers that could be the key insight
    • Models always favor the client's opinion
    • Models only summarize the first document
    • Models refuse to summarize interviews
  3. Waa maxay habka ugu nabdoon ee loo isticmaalo AI falanqaynta tirada, sida uu dhigayo hagaha?
    • Accept numbers stated in the chat reply
    • Ask the model to estimate from memory
    • Have the AI write code or formulas you can inspect and rerun
    • Average answers from multiple chatbots
  4. Waa kuwee dhaqanka ku habboon talada sirta ee hagaha?
    • Paste client data into any free chatbot if names are included
    • Ignore client rules if the firm approves the tool
    • Assume all AI tools delete data immediately
    • Use firm-approved tools whose terms exclude training on customer data and follow client rules
  5. Jiilka la kordhiyay, maxaa dhacaya ka hor inta uusan tusaaluhu ka jawaabin?
    • The model retrains on all firm documents
    • Relevant document chunks are found by similarity search and placed in the model's context
    • A human manually types the answer
    • The model searches only its training data
  6. What long-context weakness does the technical section mention?
    • Models cannot read PDFs
    • Models always ignore the first sentence
    • Models can pay less attention to material in the middle of long inputs
    • Long inputs make models refuse to answer
  7. What apprenticeship concern does the guide raise for junior consultants?
    • If first-draft groundwork shrinks, juniors may have fewer ways to build judgment
    • Juniors will be required to code in assembly
    • Juniors will no longer meet clients
    • Juniors will only format slides
  8. What slide convention does the guide say AI can help with in storylines?
    • Using as many bullet points as possible
    • Placing conclusions only in footnotes
    • Removing all headlines
    • Action titles, where each slide headline states the conclusion