quiz bu lëkkale ak tegtal · Dëgër Niveau
IA ngir laaj ak tontu ci konsultan yi
Dafay saytu gis-gis bu jagged frontier bi, naka la synthese buñ yokk ci seet mënee lajj, ak sàrti kumpa ngir dugal done kiliyaan yi ci jumtukaayi IA.
Laaj 1 de 8
Lan la Daara Business bu Harvard ak BCG ci 2023 leeral?
Xoolaatal laaj bu nekk8 laaj
Sooy def quiz bi, tontu yi ak leeral yi dañuy feeñ.
- Lan la Daara Business bu Harvard ak BCG ci 2023 leeral?
- 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
- Lan moo waral guide bi di artu ni synthese gëstu IA nekkul neutral?
- 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
- Lan mooy anam wi gëna wóor ngir jëfandikoo IA ngir jàngat limu, buñu sukkandikoo ci tegtal bi?
- 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
- Ban jëf moo méngoo ak li guide bi digle ci wàllu kumpa?
- 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
- Ci jikkoom-yokkum jëmmal, lan mooy xew balaa model bi di tontu?
- 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
- 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
- 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
- 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