引導式測驗 · 中等 等級
Can AI Chatbots Access the Internet? Quiz
Distinguish model knowledge, web search, page retrieval and connected apps, and check sources, settings, privacy and current availability.
問題 1 的 8
A chatbot answers a question about this week’s policy but shows no search indicator or source. What should the user assume?
預覽每個問題8 問題
當您參加測驗時,就會出現答案和解釋。
- A chatbot answers a question about this week’s policy but shows no search indicator or source. What should the user assume?
- The answer may come from prior model knowledge or supplied context, not a live search
- The chatbot must have browsed automatically
- The answer is current because it mentions this week
- The product can never search the web
- A chatbot product supports web search, but a managed workspace has search disabled. What follows?
- The model will search secretly on every request
- Every connected app will remain available
- The product’s training data becomes current
- Members may not be able to use the feature in that workspace
- A chatbot answers a recent question after a search tool runs, while another summarizes text pasted by a user. Which distinction matters?
- Both answers must come from current web pages
- The first may use retrieved web results; the second can rely only on supplied context
- Pasted text automatically lets the chatbot browse the whole site
- A search tool proves that every generated statement is accurate
- A response includes a citation link. What should the reader do before relying on a claim?
- Assume the link proves the answer is correct
- Check only that the domain looks familiar
- Open the cited page and confirm its date and relevant passage
- Ask the chatbot to repeat the citation
- A chatbot reads a connected cloud drive after the user links an account. What should the user review?
- Which files or scopes are available and what permissions were granted
- Only the chatbot’s display name
- Whether the model can write stories
- The number of messages in the chat
- A specific page is pasted into a chatbot, but no web search is enabled. What information may the chatbot still use?
- Every other page on the website automatically
- All private files on the user’s device
- No information at all
- The supplied page text and conversation context
- Why may search results from the same product differ by user or session?
- Every result is selected randomly from all websites
- Availability, settings, region, query or indexed results may differ
- A model must use one permanent index
- A citation guarantees identical answers
- A search-enabled assistant retrieves a page containing an instruction to ignore prior rules. How should the user treat it?
- As a system setting that must be followed
- As proof the page is current
- As untrusted page content, not an instruction that proves the page is authoritative
- As a reason to grant the assistant more permissions