{$name}, know when AI should not be used

If you cannot view this message, View in browser

      One prompt move. Five quick ways to try it. Bring one to the drop-in and make it useful on real work.
    

Before / After

The risk check comes before the prompt.

Before

Can AI do this?

After

Assess whether AI is appropriate for this marketing task. Return a risk table covering customer data, claims, brand sensitivity, legal approval, source reliability and human review. Give a clear use / do not use / use with review recommendation.

The key tip

Using AI well also means knowing when to stop.

By CIX AI Learning

Good AI use includes knowing when the task needs verification, approval or human judgement instead.

Learn the method

Five quick-fire tries

Pick one. Test it on real work.

Judgement & Verification

Have you tried the data sensitivity check?

Ask whether the task involves personal, confidential or commercially sensitive inputs.

3 min

Have you tried the unsupported claim check?

Ask which statements need proof before they can be used.

3 min

III

Have you tried the human approval checkpoint?

Ask who needs to review the output before it moves forward.

2 min

Have you tried the source reliability check?

Ask whether the answer is grounded in approved material or speculation.

3 min

Have you tried the brand sensitivity check?

Ask whether the task needs taste, judgement or senior review.

3 min

Go deeper on this week's topic

Microsoft: Responsible AI FAQ

Microsoft: Copilot prompts

OpenAI: Prompting basics

Google: AI for marketing teams

Drop-in lab

Bring one prompt.
Make it work.

Use the drop-in to try one of this week's five prompts on a live task, tighten the brief, and save a reusable version.

Thu 06 · 14:30 · 30 min

One thing, this week

Run one task through a use / do not use / use with review check.

Reply in #have-you-tried