Practice set: what assistants are, good fits, invention and risk

P1. An AI assistant’s reply is best described as:

• Generated text that is likely given your request
• A guaranteed fact
• A database lookup
• A legal opinion

P2. Which task is usually a good fit?

• Brainstorming three subject lines
• Quoting a law with its clause
• Calculating tax on 40 invoices
• Reporting yesterday’s news

P3. A precise statistic with no source is:

• A warning sign
• Proof
• A style choice
• A heading

P4. The four verification checks include:

• Source, specifics, sense and stakes
• Colour, font, length and tone
• Speed, cost, size and brand
• None

P5. High stakes and hard to check means:

• Do not rely on the assistant alone
• Use it freely
• Skip the task
• Ask it twice

P6. Who is accountable for what you send?

• You
• The assistant
• The vendor
• Nobody

P7. Summarising a report you have not read is risky because:

• You cannot spot omissions or inventions
• Summaries are illegal
• Reports are short
• The tool is slow

P8. Confidential information in a task may:

• Rule out a public tool whatever the grid says
• Make the task easier
• Be ignored
• Improve the answer