Over the past few years, we have mainly used artificial intelligence to produce more. Perhaps its most valuable application is not helping us generate more, but helping us recover the time and attention we need to understand people better.
It is probably not the first time you have seen an image, a video or a campaign and, before understanding what it is trying to tell you, thought: “This is AI.”
A few years ago, spotting it was part of the novelty. We noticed impossible hands, strange movements or that kind of perfection which, precisely because it was too perfect, stopped feeling real. We shared it, laughed about it and tried to figure out how it had been made.
Now the feeling is different. Artificial intelligence is no longer an occasional surprise or a trend. It has become a constant in the industry. And when we recognise it in a campaign, our reaction is not always curiosity. Sometimes it is distance, fatigue or the feeling that the tool has ended up taking the place of the idea.
When AI Becomes the Message
In December 2025, McDonald’s withdrew a Christmas advert in the Netherlands that had been produced using generative AI after it received a negative response. The campaign portrayed the festive season as a succession of chaotic situations and invited people to take refuge in a restaurant until January.
The criticism was not only directed at the visual inconsistencies in the generated images. People also questioned the tone, the lack of warmth and the distance between the message and what many people associate with Christmas.
Some time earlier, Google had removed the Dear Sydney campaign from its Olympic television rotation. The advert showed a father using Gemini to help his daughter write a letter to athlete Sydney McLaughlin-Levrone.
In this case, the controversy was not caused by a poor image or an impossible movement. The question went deeper: why delegate the most personal part of the experience to AI?
They are two different cases. In the first, the technology ended up overshadowing the story and the brand’s identity. In the second, it was placed in a space where effort, imperfection and a personal way of expressing oneself were an essential part of the message.
Neither case proves that a campaign will perform badly simply because it uses artificial intelligence. But both force us to ask an important question:
Is the problem that people notice the AI, or that we have not thought carefully enough about the role it should play?
Tests conducted by NielsenIQ with AI-generated advertisements point in this direction. Participants spontaneously identified many of the generated pieces and rated them as more boring, annoying or confusing than conventional advertisements.
However, the same research suggested that these tools can be useful in the early stages of the creative process, for example when preparing storyboards, exploring ideas or developing brand assets before final production.
The difference, therefore, is not simply whether or not we use AI. It is knowing what for, when and with what intention.
My Work Has Shifted
I work as a full-stack designer at DeMomentSomTres, specialising in web design and user experience. This means that a large part of my job involves understanding what people need, turning those needs into a digital experience and making sure that design decisions work beyond the Figma screen.
I have always felt comfortable experimenting with artificial intelligence. I have never seen it as something external to design or as a threat that should be kept away from the creative process.
But if I compare the way I worked three or four years ago with the way I work today, it is clear that my job has changed.
Today, we can get to a first testable prototype sooner. We can work with temporary copy that is more realistic than the usual lorem ipsum. We can generate provisional visual materials when the final photography or assets do not yet exist. We can explore more alternatives, filter options and test a hypothesis before investing in final production.
This does not mean delivering a draft as if it were a finished solution. Nor does it mean allowing a tool to decide the experience, the identity or the direction of a project.
It means something much more useful: we do not have to wait until everything is final before we can start learning.
Instead of spending many hours polishing a version that we do not yet know will solve the problem properly, we can reach a sufficiently coherent prototype sooner and observe what people understand, where they hesitate, what they cannot find and what we need to change.
It is not necessarily about finishing sooner. It is about reaching the point of validation sooner.
And that allows us to spend more time on the things that still determine the quality of the result:
Understanding what the client is really asking for. Framing the problem better. Preparing meetings with more context. Testing with users. Analysing their reactions. Identifying contradictions. Reviewing accessibility. Deciding what is unnecessary. And, above all, knowing which options should not be developed any further.
When generating a first version becomes easier, judgement does not become less valuable. It becomes more valuable.
AI can generate more answers. My job is still to know which questions are worth asking.
What Do We Do With the Time We Get Back?
In 2024, a reflection by writer Joanna Maciejewska became widely shared. In essence, she said that we wanted AI to do the laundry and the dishes so that we could create, not to create so that we could keep doing the laundry and the dishes.
The idea captured the feeling of that moment very well. It seemed that we had started automating precisely the activities we associate with expression, identity or enjoyment, while repetitive tasks remained exactly where they were.
Now, however, the conversation can be more nuanced.
Artificial intelligence can also play a role in creative processes. It can help us explore a direction, unlock a first version, test a structure or build provisional material.
The boundary is not as simple as separating creative activities from mechanical ones.
Perhaps the real difference is something else:
Què podem delegar sense perdre valor i què, quan ho deleguem, perd precisament el seu sentit?
AI can help prepare a first draft. But that does not necessarily mean it should write a young girl’s letter to her sporting role model.
It can generate alternatives for an interface. But it cannot take responsibility for deciding which option best respects the needs, limitations and context of the people who will use it.
It can speed up part of the process. But we still have to decide what we do with the time we recover.
Taking the Screen Out of the Way
Think about a medical consultation.
For years, a significant part of the conversation has taken place with a screen in between. The healthcare professional listens to the patient while also having to type, review fields, organise information and make sure everything is properly documented.
Some ambient AI systems can listen to the consultation and prepare a first draft of the notes for the professional to review afterwards.
It may seem like a less spectacular use case compared with generating a campaign, a photograph or a video. But its impact can be much deeper.
A study involving 46 healthcare professionals associated the use of these assistants with a 20.4% reduction in time spent on notes per consultation and a 30% reduction in documentation outside working hours. Professionals also reported feeling more engaged with their patients.
A later study involving 263 professionals across six healthcare systems also observed improvements in documentation time, cognitive load, attention devoted to patients and professional burnout.
The professional can look at the person. Notice a silence. Observe how they explain what is happening to them. Ask a question that had not been planned.
AI does not automatically turn a consultation into a more human experience. These tools require professional supervision, appropriate protocols and review of their outputs. In fact, professionals themselves have also identified limitations related to accuracy, the length of notes and the need for editing.
But when it is used in the right place, AI can remove part of the mechanical work from the conversation so that the person providing care can be more present.
AI had not accelerated the conversation. It had removed part of the mechanical work so that the interaction could be better.
Technology Also Needs to Know When to Step Back
There is an idea in design that I find particularly interesting: not everything that creates value needs to demand people’s attention.
Sometimes a system is better designed precisely because it does not force us to think constantly about the tool. It organises information, supports a decision, reduces friction and steps back once it has done its job.
The same can happen with artificial intelligence.
Some of its most valuable applications are not necessarily the ones the end customer sees. They can work in the background:
- Organising information.
- Retrieving internal knowledge.
- Connecting tools.
- Preparing documentation.
- Supporting an initial analysis.
- Reducing manual handovers.
- Or helping us reach a version sooner that we can test with real people.
The Center for Humane Technology argues that the conversation around AI needs to move beyond a simplistic divide between promises and dangers and focus instead on how design and incentives shape its impact on society. Its current project focuses on a particularly relevant question: what should we preserve so that technology strengthens, rather than weakens, the things that give meaning to human life?
Putting people at the centre does not mean rejecting artificial intelligence. It means deciding what role we give it and what criteria we use to evaluate its results.
Not just how many pieces of content it generates.
Not just how many hours it saves.
But also what it allows us to listen to more carefully, understand better, decide better or return to doing with greater attention.
Perhaps good AI is like good design: it makes the journey easier, reduces friction and knows when to step back once it has done its job.
Maybe You Don’t Know Where to Apply It Yet
You may know that artificial intelligence can help your company, but you may not be sure where to implement it.
You may know that artificial intelligence can help your company, but you may not be sure where to implement it.
Perhaps the opportunity lies in internal documentation that is difficult to retrieve. In a process that requires copying information between different systems. In questions that are repeated every week. In an early prototyping phase. In meeting preparation. In sales follow-up. In connecting the CRM, ERP, email and the tools your team already uses.
It may also be that the company does not need a new tool yet.
Perhaps the first step is to organise the information, review the processes and identify where time, context or decision-making capacity is being lost.
At DeMomentSomTres, this is how we approach artificial intelligence: first, we analyse real processes and needs; then we identify whether the right solution is an internal assistant, an automation, an integration, a knowledge management system or another solution tailored to the company.
Because the goal is not to adopt AI for the sake of it. It is to identify where it can reduce friction without removing precisely what we want to preserve: judgement, autonomy, creativity and human relationships.
Over the past few years, we have mainly used AI to produce more. Perhaps its most valuable application is not helping us generate, but helping us recover the time and attention we need to understand people better.
The best AI is not necessarily the most visible one, but the one that gives us back time, attention and the ability to connect.
