The Rise of the Brainstorm Bots

I was listening to The Life Scientific this morning, where the guest was Michael Wooldridge, professor of computing science at the University of Oxford, talking about AI and sentient robots. It’s a good listen; the part that stood out to me most was this…

“One of the unexpected applications of this technology is brainstorming. 

So traditionally, you know, you’re an advertising executive and you’re hired to come up with three slogans for a new banana milk drink, what do they do they lock you in a room with a flask of coffee and a packet of cigarettes. You’re not allowed out until you’ve got your three slogans for this.

But with this technology you can just press the button and say, give me a pitch for a new banana flavoured milk drink and it will just produce one and then you want another you just hit return again, give me another, and you can carry on doing that.

And I think people will rather startled that we think that’s something that’s creative. And there’s a debate to be had about whether it is actually creative. But actually the technology is extremely good. And it’s a really unexpected way in which this technology turns out to be really, really useful and there will be killer apps based around using this technology for brainstorming.”

Which is true, of course, to a certain extent.

And the technology is very good for generating endless combinations based on all the inputs it has swallowed previously as context.

I think it’s worth breaking down where brainstorming fits in any process though.

It comes at a moment where you traditionally would gather people with experience and data, and together they can interact to start producing lists of options and ideas.

That fact that you have (or should have) multiple perspectives in a room gives these ideas a better chance of covering all bases, and interesting thing emerging from the liminal space between everything you’ve thrown into the mix.

As a group, you’ll also be more likely to make informed decisions about which things you want to shortlist and why.

Meanwhile, though, I was curious as to what ChatGPT thought would be a good way to run this hypothetical brainstorm.

So I asked it this…

If I was running a brainstorm in an advertising agency for promoting a new banana flavoured milk drink, how would I set it up beforehand, who would come, how would I run the session, and who would decide which ideas to recommend?

It came back with the following…

Organizing a successful brainstorming session for promoting a new banana-flavored milk drink involves careful planning, a diverse group of participants, and a structured approach. 

Here’s a step-by-step guide:

1. Define Objectives:

Clearly outline the goals of the brainstorming session. For example, generating creative marketing ideas, identifying target audiences, or developing a catchy campaign slogan.

2. Select Participants:

Invite a diverse group of individuals from different departments within the advertising agency, including creatives, marketers, copywriters, designers, and even representatives from the client team if applicable. Ensure that participants have a mix of experience levels and perspectives.

3. Pre-session Preparation:

Provide participants with background information about the product, its unique selling points, and the target market.

Encourage participants to come to the session with some preliminary ideas, ensuring a more productive and focused brainstorming process.

4. Set Up the Room:

Arrange the seating to promote collaboration and creativity. Consider using an open space with whiteboards or flip charts for idea visualization.

Ensure there are writing materials, sticky notes, and any other tools necessary for participants to express their ideas.

5. Icebreakers:

Begin the session with icebreakers or warm-up activities to foster a comfortable and open atmosphere. This can help participants feel more relaxed and creative.

6. Guiding Questions:

Present participants with key questions related to the product, target audience, and marketing objectives. This will guide their thinking and help generate more focused ideas.

7. Divergent Thinking:

Encourage participants to brainstorm individually first, allowing for a burst of divergent thinking. Have them write down as many ideas as possible without judgment.

8. Group Sharing:

Have participants share their individual ideas within small groups. This fosters collaboration and helps build upon initial concepts.

9. Convergent Thinking:

Bring the groups back together to discuss and consolidate the best ideas. Encourage the merging of concepts and the elimination of duplicates.

10. Idea Prioritization:

Implement a method for voting or ranking the ideas to determine which ones resonate the most with the group. This could involve a simple voting system or a more structured decision-making process.

11. Decision-Making:

Depending on your agency’s structure, a designated decision-maker or a collaborative team may be responsible for selecting the final ideas to recommend. Consider factors like feasibility, creativity, and alignment with marketing objectives.

12. Action Plan:

Develop a plan to implement the selected ideas, including assigning responsibilities, setting timelines, and establishing milestones.

13. Follow-up:

Schedule follow-up meetings to track progress, address challenges, and make any necessary adjustments to the campaign strategy.

By following these steps, you can create a well-organized and effective brainstorming session that harnesses the creativity and expertise of your advertising agency team.

Now, of *course* it thinks this is the way it should be run… at some stage, it’s swallowed enough content from the past to infer that this structure, and the steps as detailed, are likely enough answers to make the reader happy.

What might be worth doing now is asking it which part of the process could be effectively supported or replaced by AI today.

Which is what I did.

Me: Of this process, what would you recommend replacing with AI?


While AI can be a valuable tool in the brainstorming and decision-making process, it’s important to recognize that certain aspects of the process are best handled by human creativity, intuition, and emotional intelligence. 

Here are some areas where AI can be integrated:

i. Data Analysis and Market Research:

AI can analyze vast amounts of data to provide insights into market trends, consumer preferences, and competitor strategies. This information can help inform the brainstorming process and provide a data-driven foundation for creative ideas.

ii. Idea Generation:

AI algorithms, including generative models like GPT-3, can assist in generating a large volume of ideas based on input prompts. While these ideas may lack the emotional and creative depth of human-generated concepts, they can serve as a starting point for further human refinement.

iii. Content Optimization:

AI tools can help optimize content, such as ad copy or visuals, by analyzing performance metrics and suggesting improvements based on data. This can enhance the effectiveness of the creative output.

iv. Decision Support:

AI can assist in the decision-making process by providing data-driven insights and recommendations. It can analyze the feasibility and potential success of different ideas based on historical data and industry benchmarks.

However, there are crucial aspects of the brainstorming process where human involvement is essential:

1. Creativity and Innovation:

Human creativity is often necessary to come up with truly original and innovative ideas. While AI can provide suggestions, the ability to think outside the box, make unique connections, and understand cultural nuances is a distinctly human trait.

2. Emotional Intelligence:

Understanding human emotions, cultural contexts, and societal trends is vital in creating campaigns that resonate with the target audience. AI lacks the emotional intelligence required for this aspect of advertising.

3. Collaboration and Team Dynamics:

The collaborative nature of brainstorming sessions relies on the diverse perspectives, experiences, and insights of human participants. Team dynamics, spontaneous interactions, and the ability to build on each other’s ideas are integral to the creative process.

4. Final Decision-Making:

While AI can provide recommendations, the final decision on which ideas to pursue is often made by individuals or teams with a deep understanding of the brand, the market, and the overarching goals of the advertising campaign.

In summary, AI can be a valuable support tool in specific aspects of the advertising brainstorming process, particularly in data analysis and optimization. However, the core creative and human-centric elements of ideation and decision-making are best handled by the creative minds within the advertising agency.

OK, fun experiment. But what might it mean?

I think, going back to the original quote from Michael Wooldridge, it’s easy to pick one particular aspect of a process out of the system it lives in, and imagine ‘how AI might replace exactly this’.

It ignores the relationships between all the other parts of a process though, and therefore in practice gives you a whole set of other problems to solve.

Clearly, AI can generate lots of material really quickly. And yet there comes a limit at which you need to think about when is there too much for the people in this loop to process?

Editing the outputs, and selecting what you want based on the context you’re operating in, becomes a different sort of human skill, and one that we might even have to reimagine the education system to start instilling in people.

And finally, you can by and large make an AI system like this say whatever you like with the right prompting. That moment of inception is the vital bit… who decided that the example should be a banana floured milk drink in the first place?