引導式測驗 · 中等 等級

使用 ChatGPT 進行財務建議測驗的風險

測試 ChatGPT 財務範圍、連接資料限制、財務記憶、來源驗證、隱私以及建議或交易限制。

問題 1 的 8

An eligible U.S. user linked a bank account in ChatGPT Finances. Which task do OpenAI’s current docs support?

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。

  1. An eligible U.S. user linked a bank account in ChatGPT Finances. Which task do OpenAI’s current docs support?
    • Move funds between connected accounts after the user asks in chat
    • File the user’s tax return using synced transaction data
    • Review spending and recurring bills from connected data, while checking its source and last sync
    • Trade securities automatically when portfolio allocation changes
  2. If an account is not connected and a user has not shared their financial details, why might a response fail to fit their situation?
    • The answer is necessarily tailored to the user’s jurisdiction and filing status
    • It may lack goals, time horizon, taxes, risk tolerance, or jurisdiction
    • It receives verified account and tax records automatically
    • It determines suitability from current market prices regardless of personal goals
  3. What should a user do with a tax threshold from an answer?
    • Check a prior-year tax table and assume the threshold is unchanged
    • Confirm the number in a current secondary article without opening its references
    • Accept a citation if the generated answer includes a quotation
    • Verify it with a current primary source
  4. Which information should a user avoid pasting manually into a general-purpose chat instead of using an approved account-connection flow?
    • Passwords, account numbers, tax identifiers, and confidential documents
    • A fictionalized example that contains no real account details
    • A public IRS publication relevant to the question
    • A round-number budgeting hypothetical with no real account or person identifiers
  5. Does a confident response establish that a strategy is safe?
    • Yes, because a detailed risk list proves that the strategy fits the user
    • Yes, when several generated answers reach the same conclusion
    • No; fluency does not prove accuracy or suitability
    • Yes, if the model’s training data include historical market information
  6. How should users verify model-provided citations?
    • Open the source and check that it supports the claim and is current
    • Trust the generated citation when its publisher is reputable without checking the cited passage
    • Check only that the link title resembles the answer’s wording
    • Rely on a citation when its domain looks like a financial regulator
  7. For a consequential personalized choice, what can a financial professional add to a model explanation?
    • Provide an opinion based only on a model summary without checking personal details
    • Evaluate the decision using the user’s actual circumstances
    • Make a recommendation without considering the user’s goals or constraints
    • Take over the user’s authority to set financial goals
  8. Can ChatGPT guarantee investment results?
    • Yes, if the portfolio is diversified and rebalanced
    • Yes, if the answer uses recent prices and market history
    • No, but it can ensure that future investment fees will be lower
    • No; markets and future outcomes are uncertain