Author: martina.monterisi

Starting with a problem statement

Data visualization is a field built on making information accessible, engaging and clear. AI is now present at every stage of that process, but almost no one in the field has a shared framework for how to properly use it. Most creatives are learning by trial and error, which in practice means learning by wasting: wasted prompts, wasted time, outputs that miss the mark because the input was never quite intentional. For newcomers, the barrier is even higher, since the technical language is intimidating and there is no obvious starting point.

Beyond individual frustration, the wider feeling of uncertainty is something real. Unstructured AI use in storytelling can reinforce structural bias, flatten narrative diversity, and quietly shift creative authorship away from the human without anyone deciding that should happen.

The Solution

The methodology this project is building is a practical open framework that guides creatives through the stages of data-driven storytelling, with clear indications of where AI genuinely helps, how to write prompts with purpose, and how to stay in control of the narrative throughout. It can be described as a documented design thinking process, designed to be used as a reference, taught in academic contexts, or adopted by studios building internal guidelines.

It works by breaking the creative process into stages, assigning AI a defined and intentional role at each one, and giving users the vocabulary and structure to make decisions rather than just react to output.

The target audience is anyone working at the intersection of data and narrative: visualization designers, data journalists, researchers, students, and freelancers. The customer, in an academic and institutional sense, could be universities, design programs, and creative organizations looking for a responsible framework to teach or reference.

The change is not dramatic. It looks like a field that slowly develops a common language for something it is already doing.

Should we really talk about money?

Honestly, monetizing this personally feels like the wrong frame for what it is. But if we should consider this option, there are a couple possible paths worth naming. Institutional licensing to universities or design schools that want to integrate the framework into their curricula is the most natural fit. Funded research continuation through academic grants is another. Further down the line, a workshop or short course format built around the methodology could generate income without compromising the open-access nature of the core framework. The goal is not actually profit, yet reach.

Here follows a possible business model structure that could work for such idea.

Every research project, at some point, has to find an answer to the following question: who actually needs this, and why would they reach for it? Mapping out the customer profiles and value proposition for this methodology felt like one of the more revealing steps of the process, mainly because it forced me to stop thinking like a researcher and start thinking like someone who has to explain why this matters to a stranger. Two potential users came to mind immediately, and they arrive at the same problem from different starting points.

Profile 1: the intermediate creative or researcher

This person already works in data visualization or data journalism. They know their tools, they have communicative goals, and they have probably already experimented with AI in some part of their process. The frustration is not unfamiliarity. It is the lack of structure. They write prompts without a clear framework, use AI across too many stages, and spend more time correcting output than creating. What they want is a workflow they can repeat, trust, and call their own. A process that saves resources, reduces noise, and keeps their authorial voice intact.

Profile 2: the student or freelancer new to data-driven storytelling

This person has just entered the field. They might be studying design, communication, or journalism, or picking up freelance work that is pushing them toward data narratives for the first time. AI feels both exciting and overwhelming. They do not yet have a reference point for what “good” looks like in this process, and the specialized terminology alone is enough to make them feel like they do not belong. What they need is not just a workflow but something that builds their confidence and knowledge at the same time as it guides their practice.

What this methodology could offer to both

For both profiles, the core offer is the same: a step-by-step framework that makes intentional AI use in data storytelling accessible, documented, and repeatable. For the intermediate user, it brings order to an already active practice. For the newcomer, it lowers the barrier to entry without oversimplifying the field.

The pain relievers are practical: fewer wasted prompts, a clear structure for each stage of the creative process, a built-in glossary so no one has to go looking for definitions elsewhere, and a simpler visual version for those who find dense theoretical language a barrier.

The gain creators go deeper. Both profiles walk away with more than a finished project. They build transferable skills, develop a personal voice in how they collaborate with AI, and become more conscious and ethical users of a tool that is not going away.

Breaking down the accessibility requirements of this project felt like one of the more hard exercises of the whole research process, because we are used to think everything would have been “easy and smooth” since it’s a niche digital research product. But there was much more behind.

What the user should be capable of

On a physical and personal level, sight would be the most relevant sense, since the experience is largely text-based, but it can be made compatible with screen readers and text-to-speech tools from the start. Hearing is not a requirement at all. Movement needs are minimal too, limited to basic typing, clicking, or scrolling, with voice dictation and keyboard-only navigation as fallbacks.

On the cognitive level is where it gets more demanding. Following an iterative loop of prompting, reviewing, and editing requires sustained mental focus and a willingness to sit with a process that is genuinely not instantaneous. As well as the need for a strong digital literacy to navigate AI tools or terminology, and enough language confidence to work in what are predominantly English environments. Though nowadays everything can be translated in real time.

Financially, the core methodology is designed to be of course a free to access tool, to be used as a starting point or guideline to use AI tools. Infrastructure needs are also few: a device and internet connection.

Who is it meant for and where does this happen

The methodology will be designed for anyone engaging with storytelling on a professional or exploratory level, from researchers and creatives to students and private users. It lives entirely in digital space, which means it can happen in any place and environment.

What it does require is probably the mindset: willingness and awareness to co-create with a machine, and enough critical thinking to question what the machine gives back.

The barriers

Two barriers kept coming up in the entire thinking process. The first was language, since AI tools lean heavily on English, nuance is often the first casualty of translation. The solution here could be to develop a simplified and even more visual version of the methodology that relies on basic English, diagrams, and examples rather than dense theoretical language.

The second barrier was knowledge. The informatics specific terminology involved is genuinely intimidating and not of common knowledge (it was for me too). The solution I thought for this issue was to add a vocabulary directly into the framework itself, so all the basic knowledge can be in the same place as the product.

To conclude, cognitive overload is also worth a mention. When the prompting loop feels endless and the output feels overwhelming, the step-by-step structure of the methodology becomes less of a nice-to-have and more of a way to escape.

After outlining the actors and affected people and sectors within a possible new direction in the creative process, let’s try to visualize what the current state of things look like and what what it could become after the introduction of a new methodology.

Before

• involvement of AI in all stages of storytelling generation/ideation/creation
• no precise and documented knowledge on how to do it in this precice field
• the prompts to ai tool not well written, without puropose
• Too many requests = extreme waste and environmental impact
• Unethical use of AI

After

• easy to use / step by step methodology to understand and use a correct workflow in the creation process
• Aware and ethical use of AI tools
• Less prompts / requests sent with more efficient answers
• Open a new dialogue
• Set a standard or starting point for the field

Right now, most creatives are figuring this new directions out alone. A structured methodology could change that and optimize our workflow without making us feel left out of the process. It’s not just about better prompts, yet it’s about reflecting on how an entire field relates to a tool that is now part of our everyday life.

Trying to make sense of how AI fits into the creative process meant also looking first of all at who is actually involved in this process and who could be potentially affected by something new in this field. The system map I put together tries to capture the messy, layered connections that a new methodology for AI-assisted storytelling would actually have to deal with. What surprised me most has been how quickly the actors in the circles can expand and increase. The part I keep coming back to is where genuine human intentionality sits in all of this. When so many actors are pulling in different directions, tracing where creative agency actually lives feels less like a technical problem and more and more a human one.

After starting with a giant question mark about the role played by AI nowadays in the field of data visualization, I decided to start with a literature review to narrow down and frame better on which topic/s to focus. So I have decided to investigate better how LLMs and generative models intervene in data driven storytelling (meant as turning data into easy-to-read and easy-to-understand stories that help turning insights into action).

A systematic review regarding telling stories with data deeply guided my curiosity and focus on the topic of human intentionality. In addition, many papers were also questioning the modern role of the author in the creative process. I want to understand better where the author exactly stands today and how empathy and human intention, that are not entirely replicable by machines, fit in a world dominated by algorithmic storytelling. Therefore, I am exploring the co construction of meaning to observe how humans and algorithms might merge to build these new narratives.

As my literature review expanded, several critical points have emerged. I’d like to explore how algorithm suggestions might perpetuate a structural bias compared to a possible unintentionally “human manipulation”. It is also crucial to question whether relying on automated micro narratives is always the right choice when considering a diverse user base. To conclude (as if this was not already enough qeustions) I plan to explore the sociological aspect of the epistemic control, understanding to whom may it belong.

But for now, let’s keep the focus on just the creative process adopted to create a narrative starting from data and see how the data community behave when involving AI!