Exploring Generative AI’s Impact in Creative Ideation

As part of our ongoing exploration of AI’s evolving role in creative work, we recently hosted a Holition Plays workshop titled 'Is Generative AI the New Undo Button in Creative Thinking?'

The title was inspired by John Maeda’s Laws of Simplicity, where he reflects on how the shift from physical to digital design introduced the "undo" button, a simple feature that fundamentally changed the creative process. While it gave designers more freedom to experiment and take risks, it arguably also reduced the depth of engagement with their ideas, as the safety net of "undo" made decisions feel less final.

As an agency that has been experimenting with different forms of AI for decades, and more recently, extensively with Generative AI for creative visualisation, this workshop was a natural next step. It built on earlier internal sessions, including a workshop on prompt crafting for image generation. But this time, rather than focusing on outputs, we wanted to zoom in on the process itself and the deeper human and cultural dynamics at play. Inspired by Digital Anthropology, we explored how technology, creativity and human behaviour intertwine in the act of ideating with AI.
 

We posed a series of questions to guide the session:
- How does Generative AI influence the variety and depth of ideas that are created?

- How does ideation differ in a group environment with and without Generative AI?

- How does Generative AI impact authorship and ownership?

- And ultimately, what are the long term and short term implications for our creative confidence as AI becomes part of our thinking process?

Setting the Context with Thought-Provoking Talks

We kicked off the evening with a presentation by Anita Benko, our Head of UX and Strategy. Anita raised key questions about AI’s role in the earliest stage of ideation, when the blank page stares back and the familiar wave of creative panic sets in. Is this moment of discomfort an essential part of the human creative process? Does using AI to bypass that state limit or expand the depth of our thinking? And, as our interactions with AI shift from occasional to constant, will our sense of creative confidence, our self-efficacy, become increasingly rooted in our AI augmented selves? Will reliance on AI reshape not only how we create but how we perceive our own creative capability?

This was followed by a presentation from Dr. Eleonora Pantano, Associate Professor in Retail and Marketing Technology at the University of Bristol, who shared insights from her research on how Generative AI-designed luxury brand assets, specifically luxury bags, are perceived by consumers, both when they are aware and unaware that the design was AI-generated.

Her research highlighted a crucial shift: as AI design tools become more accessible and more widely used, the product’s design may no longer serve as the core differentiator. Instead, the true competitive edge will increasingly lie in the quality of the manufacturing process and the materials used.

In other words, while Gen AI opens the door to a new level of creative opportunity, making it easier for both luxury and non-luxury brands to replicate sophisticated design language, it also shifts the balance of brand value away from aesthetics alone.

The final talk of the evening came from Alina Arefyeva, Head of Creative at Holition, who shared her personal experience of integrating Generative AI into her creative workflow. 

Alina highlighted an important caveat: working with AI requires the creative direction to be defined much earlier in the process. Because AI lacks the intuitive human understanding of style and context, the designer must be precise and intentional with prompts from the start, essentially forcing an earlier conceptualisation of creative instincts before generating visual outputs. She also pointed out the importance of maintaining focus during the process.

With the fast and high end generation of images, it is easy to get distracted by refining visuals to look like finished products, rather than stepping back to understand why those visuals exist in the first place and leaving space for improvisation, exploration and interpretation.

Splitting Participants: AI vs. Non-AI Groups

Following the presentations, participants were split into two groups to work on the same creative brief: one group was encouraged to incorporate Generative AI into their process, while the other worked entirely without it.

Inspired by Eleonora's research, the creative brief invited everyone to imagine ways of extending the concept of a bag, whether a handbag, clutch, briefcase, backpack or otherwise, by adding new functionalities that address contemporary needs. 

Rather than focusing on aesthetics or style, the emphasis was placed on rethinking the function of the bag to make the exercise accessible and inclusive for participants from a wide range of backgrounds.

We referenced examples such as Ray-Ban and Meta’s AR glasses (whether or not you’re a fan of the idea), Carolina Herrera’s lipstick key rings, Coperni’s Tamagotchi Bag, and fingerprint-protected security bags, to help shift thinking beyond traditional features. This setup allowed us to observe not only the types of ideas each group developed, but also how the creative process itself unfolded, from the initial brainstorming phase to the way concepts were refined.

Observations from the Workshop

Working With vs. Without Gen AI
Interestingly, more participants chose to work without Gen AI, despite the workshop’s aim to explore its creative potential. Dr. Eleonora Pantano explains that this decision often stems from an underlying skepticism, particularly when the task is creative. 

Many voice an unconscious fear that AI might "steal" or diminish a trait often viewed as distinctly human: the ability to generate original, imaginative ideas. In contrast, when it comes to non-creative or more technical tasks, there is less hesitation about using AI tools.
 

How Did Participants Collaborate with AI?
Across the board, AI was rarely used to generate the initial idea. Participants preferred to start with their own thinking first, and only once some direction was formed did they turn to AI to help expand, vary, or refine their concepts.

This pattern suggests a consistent view of AI not as an instigator, but as a collaborator, that can accelerate iteration or suggest alternatives, but still relies on human framing and intention. The dynamic echoed that of traditional group ideation, where the exchange of ideas prompts new angles and refinements, and where AI was simply another ‘voice’ in the conversation rather than the starting point.

The Role of Human Experience in Idea Generation
Among the non-AI groups, ideas were strongly rooted in personal experience and empathy. Their starting points often reflected real-world needs, frustrations, or observations:
- One group tackled the problem of physical strain from carrying heavy bags, exploring solutions for redistributing the load.

- Another imagined a multifunctional bag designed to relax commuters after a long day, especially for those using the London Underground.
- Other groups leaned toward security concerns, brainstorming anti-theft features inspired by life in a busy urban setting like London.

Even in the AI-assisted groups, human experience remained the foundation. For example, one team started with the idea of designing a bag for people with disabilities, and then used AI to develop more specific functionalities, such as a bag that could alert both the user and bystanders in the event of an impending epileptic seizure.

AI’s Influence on Solution Types
When AI was involved, the ideas tended to lean more heavily toward speculative, tech-driven solutions. Examples included:
- A bag that changes its appearance based on mood or weather.

- Built-in USB charging ports.

- GPS-based anti-theft systems, and even concepts involving electroshock deterrents for would-be thieves.

They also mentioned that the AI generated a long list of outputs, and it was their job to filter through it and find the relevant ones for their context. This raised an important observation: the quality and relevance of AI-generated ideas is highly dependent on the clarity and precision of the prompt.

Some groups noted that when it came to visual or aesthetic considerations, AI’s suggestions often felt uninspiring or off-target. This observation was echoed in Dr. Eleonora Pantano’s research, which highlighted the importance of providing AI with well-defined references and clear criteria, for example, feeding the system specific examples of “luxury bags” in order to guide its outputs toward desirable results.

This point was further reinforced by Alina Arefyeva, who emphasised the need for designers to clearly conceptualise and articulate their visual direction before involving AI, as the technology lacks the intuitive, shared understanding of style and context that humans naturally bring to the table.

Authorship and Originality
The workshop also surfaced rich discussions around authorship and creative confidence. Interestingly, even groups that did not use AI reflected on the origins of their ideas, acknowledging how heavily their thinking was shaped by past experiences, cultural references, and external inspiration.

This observation blurs the distinction between human and AI creativity: both rely on remixing and reinterpreting existing information. The question “Are any ideas truly original?” remained open, prompting participants to reflect on creativity as a process of iteration and recombination, regardless of whether the collaborator is human or machine.

Ownership of Ideas
When it came to ownership, the groups largely agreed: while AI can help generate ideas, it is still the human creator who makes the final decisions, refines the concepts, and assigns meaning to them. In that sense, ownership remains human-led, even if the process is AI-supported.

However, the conversation quickly moved beyond the workshop scenario to more complex, real-world questions. For example: if an AI tool like MidJourney produces a design based on a human-written prompt, who owns the output? The person who wrote the prompt? The developers of the AI? The company that trained the model?

We also raised questions about the future of authorship as more designers and creatives train AI models on their personal styles, potentially creating a new layer of collective intelligence and further complicating the ownership of ideas. Some AI companies are even considering getting illustrators involved in the process and finding ways to give them credit.

Where Next?

The workshop left us with several questions to explore further, both in terms of creative process and the broader implications for authorship and collaboration.

One direction is to focus on AI’s evolving role as a collaborator rather than an originator:
 → What new features or design principles could strengthen AI’s ability to support — rather than replace — human creative thinking?
 → How could AI better integrate into group settings, where creative ideas are typically shaped through collective dialogue and iteration, rather than individual prompting?

Another path is to dig deeper into the questions around ownership and authorship:
 → As AI becomes more embedded in creative workflows, how will ownership of ideas be defined and protected, especially when the line between human and machine contribution is increasingly blurred?
 → How might training AI on individual or brand-specific styles challenge or reinforce existing ideas about intellectual property, creative credit, and authorship?

To participate in our Holition Explores initiatives, sign up to our newsletter and follow us on LinkedIn or Instagram for more information.