How to use AI for brainstorming that beats generic ideas

Most people ask AI for "ideas" and get exactly what that vague request deserves: generic, forgettable output. The fix isn't a better tool, it's a better prompt structure that treats AI as a thinking partner rather than a vending machine for ideas.

AI is a pattern machine. Its job is to make predictions about how you should respond based on the patterns of the prompts that you send to it.

Sending a vague prompt such as "ideas for my podcast" doesn't give AI much to work with. However, giving it specific details about the target audience, desired tone, format, and purpose will be much more helpful.

In an academic research study conducted by the University of Exeter Business School and UCL School of Management, it was found that when writers had access to AI-generated ideas, their stories were considered to be more creative, more well-written, and more entertaining, especially for people who weren't naturally creative writers.

Six prompt structures worth trying

Different problems call for different prompt shapes:

1. A "wide net" prompt asks for 40 unranked ideas split into safe, surprising, risky, and practical categories, useful when quantity matters more than polish early on.
2. A "constraint box" prompt forces ideas to fit a real budget, timeline, or team size.
3. An "opposite angle" prompt asks AI to invert the conventional approach entirely, which helps break polite groupthink.
4. A "grounded research" prompt has AI act as a researcher, cataloguing audience frustrations before turning them into idea starters.
5. A "remix" prompt pairs two unrelated industries or formats to force unexpected combinations.
6. Once a shortlist exists, a "critic" prompt turns AI into a sceptical editor that pressure-tests weak assumptions and clichés.

However, the process of doing the actual work begins once the ideas are generated through AI. 

Filtering out those ideas based on their need among the audience, their novelty, and feasibility and then filling them with evidence and experiences which AI does not know is what helps distinguish a useful idea from a seemingly plausible idea.

Company policy regarding the use of any confidential strategic document or data from customers or employees should never be used in AI software without prior review since it can be retained, examined, and even made available to vendors.