Leon Furze

Applying Generative AI like ChatGPT and Copilot to brainstorming is one of the most touted uses, and there are plenty of situations where the technology is well-suited to idea generation. At the same time, it’s not always a great idea to offload the creative process to GenAI. In this post, I’ll explore a few different ways to use AI for idea generation, and also suggest when it might be best to do things the old fashioned way.

Why might it be better to avoid AI?

If you’re in a social media bubble like mine, you’d be forgiven for thinking that AI is the solution to every problem. Stuck for marketing ideas? Slap some AI on it. Trying to engage students in an online course? Chuck a chatbot in there. Looking for fifty ways to build a better website/product/dream life? Ask ChatGPT for three hundred options, and cross off the ones that are rubbish.

But it isn’t always a good idea to go straight to Generative AI when coming up with new ideas. Being able to make novel, human connections between disparate pieces of information is something we’re profoundly good at, and while AI can certainly create lists of ideas at scale, it doesn’t achieve the same outcome as simply sitting down with a pen, paper, and the contents of your own head.

Brainstorming and mind-mapping encourage you to look for connections between things you already know, and expose gaps in your knowledge. AI can’t do that for you. It’s the same for students: a quick brainstorming activity will often reveal who is engaged, who has taken content on board, and who needs a boost.

In fact, despite the focus of this article, I’d recommend you don’t use AI for most brainstorming tasks. Give the process time. Allow it to be slow and messy. You might not come up with as many ideas as a chatbot, but they will be more meaningful to you, and that’s what matters. There are also some recent studies which suggest AI-creativity isn’t all it’s cracked up to be, and while using AI for creative tasks might help the least creative member of a group, it may actually hinder those who are naturally creative and curious.

Anyone can brainstorm (and no, GenAI doesn’t “democratise creativity”), but there might be times when human idea generation can be augmented by AI assistance, or times when, to get to the real problem or skills, it’s helpful to bypass the brainstorming process entirely.

When is AI useful?

Here are a few situations where AI can be a useful tool for brainstorming and idea generation:

1. When you need a large quantity of ideas quickly. While the ideas generated by AI may not always be the most creative or meaningful, chatbots like ChatGPT can produce dozens or even hundreds of suggestions in a matter of seconds. This can be helpful when you need to cast a wide net, gather a large pool of options to pick from, or simply get the creative juices flowing.
2. When you’re stuck in a rut. Sometimes our own knowledge and experiences can limit our creativity. If you find yourself recycling the same ideas over and over, querying an AI system can introduce some much-needed novelty and help break you out of familiar patterns of thinking. The AI’s suggestions may seem odd or irrelevant at first glance, but they can spark new trains of thought.
3. When dealing with topics outside your area of expertise. If you’re brainstorming ideas for a project or problem in an unfamiliar domain, AI tools can leverage their broad knowledge to suggest approaches you may not have considered. For example, a marketer trying to name a new product could get some initial ideas from Claude to better understand common naming conventions in that industry.
4. When you want to quickly identify gaps or weaknesses in a set of ideas. Once you have a brainstormed list, you can ask an AI to poke holes in it, spot potential issues, or suggest improvements. This critical analysis can help refine and strengthen the ideas you’ve generated.

The key is to view AI as a tool to enhance and complement human creativity, not replace it entirely. Use it purposefully, but don’t become overly reliant on it at the expense of your own generative abilities. The most impactful ideas often come from the unique blend of knowledge, experiences, and imagination that only human minds can provide.

Brainstorming activities with GenAI

Rather than using simplistic prompts to endless generate lists of ideas (like, “give me ten titles for a blog post about brainstorming”), I find it’s better to use some for of scaffold or structure. This is exactly the same way I teach brainstorming and mind-mapping activities. In both of my books on Writing and Reading I stress that the skill of brainstorming and idea generation needs to be explicitly taught, otherwise we end up with the dreaded “four legged spider” mind-map.

Instead of sitting around hoping that students will spontaneously create ideas from nothing, it’s useful to teach a few methods of brainstorming, connection-making, idea generation, mind-mapping, and so on.

All of these methods can also be applied to generative AI. Instead of expecting students to generate quality ideas from basic prompts, we can provide scaffolding through existing idea-generation methods.

Thinking routines

The Project Zero Thinking Routines are always my first port of call for idea generation activities. The simple, free website contains dozens of familiar activities like see-think-wonder and think-pair-share, as well as more complex lessons like circle of viewpoints. It is very easy to adapt these Thinking Routines using GenAI. In fact, because they’re so common, most of them are “understood” by GenAI: they feature in the dataset, so you can ask a GenAI application to refer to them through your prompt without requiring any further context.

Here are a few examples of how Thinking Routines can be adapted for GenAI assisted brainstorming:

1. Generate-Sort-Connect-Elaborate: This routine encourages learners to generate ideas, sort them into categories, make connections between them, and elaborate on those connections. GenAI could be used in the initial “Generate” stage to quickly produce a large number of ideas or examples related to a topic, which students could then sort, connect, and elaborate on. GenAI could also be used to sort or connect ideas which students have generated themselves.
2. Headlines: In this activity, students create a headline that captures the essence of a topic or idea. ChatGPT or similar language models could generate multiple headline options for students to analyse and discuss, helping them identify key themes and summarise complex topics succinctly.
3. Claim-Support-Question: Students make a claim about a topic, identify support for that claim, and ask questions that challenge or expand on it. GenAI could be used to generate potential claims or counterarguments, pushing students to think more deeply about their own positions and the evidence behind them.
4. Compass Points: This routine encourages students to explore various perspectives on a topic (E – Excited, W – Worrisome, N – Need to Know, S – Stance or Suggestion). AI could help generate ideas for each compass point, exposing students to a wider range of viewpoints and considerations.
5. Tug-of-War: Students explore the tensions and complexities within a topic by identifying opposing forces or ideas. Language models like Copilot could help generate these opposing viewpoints, which students could then analyse and use to develop a more nuanced understanding.
6. Circles of Action: This routine helps students consider different spheres of influence and brainstorm actions they could take within each sphere. GenAI could assist by suggesting potential actions or consequences, encouraging students to think more broadly about their own agency and impact.

Here are a few example prompts which might be used with the above:

Here is a list of ideas we have generated about <topic>. <copy/paste list of ideas>. Use these ideas as the ‘generate’ part of a Generate-Sort-Connect-Elaborate thinking routine, and complete the sort, connect, and elaborate stages.

We are generating ideas about <topic>. Ask us questions to help complete a “compass points” thinking routine about the topic.

We have completed a circles of action thinking routine on <topic> and these are some of the positions we have discussed so far <copy/paste>. Brainstorm a list of additional perspectives, actions, and consequences we may have overlooked.

Design thinking activities

Idea generation is one of the fundamental stages of design thinking, PBL, and design sprint processes. There are already many frameworks and activities used during the design thinking process which can be readily adapted to incorporate GenAI.

Design thinking activities are another area where Generative AI can be effectively integrated to enhance the brainstorming process. Many of the common techniques used in design thinking, such as “How Might We” statements, analogous inspiration, and SCAMPER, can be augmented with the help of AI tools. Here are a few examples:

1. “How Might We” (HMW) Statements: HMW statements are used to frame problems as opportunities and spark creative problem-solving. For example, instead of saying “The checkout process is confusing,” you might ask, “How might we make the checkout process intuitive and seamless?” GenAI could be used to generate a variety of HMW statements based on a given problem, helping teams explore the challenge from multiple angles.
2. Analogous Inspiration: This technique involves looking to other industries, nature, or unrelated fields for ideas that could be applied to the problem at hand. For instance, a team designing a new organizational structure might look to how ant colonies or neural networks function. ChatGPT could be prompted to suggest relevant analogies or to help map insights from the analogy back to the original challenge.
3. SCAMPER: This technique involves examining an existing idea or product through various lenses (Substitute, Combine, Adapt, Modify, Put to another use, Eliminate, Reverse). GenAI could assist by proposing ideas for each SCAMPER category, pushing participants to consider approaches they may not have thought of on their own.
4. Empathy Mapping: This activity helps designers understand the needs, thoughts, and feelings of their users. AI could be used to generate potential user personas or to suggest additional considerations for each quadrant of the empathy map (Says, Thinks, Does, Feels).
5. Storyboarding: Storyboards are used to visually represent the user journey or the sequence of events in a proposed solution. GenAI could help generate ideas for individual frames or suggest alternative paths for the story to take.
6. Rapid Prototyping: In this phase, teams create quick, low-fidelity prototypes to test their ideas. AI could be used to suggest potential materials, features, or interactions to include in the prototype, based on the team’s goals and the insights gathered from earlier stages.

Here are a few example prompts that could be used with these activities:

We are brainstorming solutions to the following problem:. Generate 3-5 “How Might We” statements to help us reframe the problem and spark new ideas.

We are looking for analogous inspiration to help us innovate on. Suggest a few examples from nature, other industries, or unrelated fields that could provide relevant insights or ideas about <topic>.

We are using the SCAMPER technique to innovate on. For the S and P sections of the SCAMPER, suggest potential substitutions or alternatives we could consider:

We are creating an empathy map for. Based on the characteristics we’ve identified so far, what are some additional thoughts, feelings, or behaviours this user might have?

We are storyboarding the user journey for. Here is the current flow:. Suggest 1-2 alternative paths or additional frames that could enhance the user experience.

As with the Thinking Routines, the key is to use AI as a supportive tool in the design thinking process, not as a replacement for human creativity and critical thinking. The AI’s suggestions should serve to inspire new ideas, challenge assumptions, and encourage participants to approach problems from different angles.

If you’d like to get in touch to discuss consulting and professional learning services for Generative AI, including how to incorporate activities like these across any subject area, use the form below to get in touch: