R076 - Model Designers, with Paz Pérez

The last two years have been truly crazy. If we think about the explosion of generative artificial intelligence, the speed at which models evolve, and the impact they have on our work, it's something we've never seen before. They have already changed the way we work, but who designs these models? And I don't just mean interaction design or how we interact with them, but how the models themselves are designed.
Hallucinations, privacy, AI biases are not just problems encountered by users, but are deeply rooted in the way they are built. What opportunities are opening before our eyes?
What is the role of Model Designers in creating ethical and responsible AI models? Today we talk with Paz Pérez, Senior UX Designer at Google.

R076 - Model Designers, with Paz Pérez

What is the real world to you?

Well, the real world is how each of us experiences our life, our day-to-day. The real world is how we go through our day, with our learning, our background, our things, our context —we're going to talk a lot about context— and I believe the real world is very personal in the end. Each person lives it from their own perspective, but what connects it all is how we learn and grow along the way.

I believe we are constantly learning and adding up. Each experience, each situation brings us something new. Even when we fail, there is learning. It's what makes our lives so unique and personal, but it's also what connects us with others: we are all in a constant process of learning and adapting.

You come from architecture. How did you make the leap to UX design?

Yes, I worked for many years as an architect, a couple of them here in Barcelona, which were truly wonderful years, and many in New York as well. There in New York, I was designing public schools. But I always had an eye on technology since I finished my degree. When I finished architecture in Seville, I decided to move to Barcelona and that's when we started an e-commerce in 2005-2006, when setting up sales platforms wasn't like it is now. You did everything from scratch. My partner —now my husband— and I learned a lot about technology until 2008 came and the crisis started. It was a difficult time for architecture. The crisis was a turning point, and although I continued working for several years as an architect, I never lost sight of the world of technology.

The mindset you have when designing a building, a chair, or a digital product is the same: solving design problems.

Although I continued working as an architect for several years, I was always looking at what was happening in the world of technology, which already caught my attention. There came a time when I decided to take the plunge. Architecture gave me a great foundation because designing a building, a chair, or a digital product involves the same mindset: solving design problems. The transition was natural because both disciplines require a structured approach, user understanding, and attention to detail.

And how is it working at Google and living in San Francisco?

It's fantastic. For me, the best part is the people. From the outside, you see the perks —like massage rooms or free snacks— but it's really the people who make the difference. You work with excellent professionals. Also, Google greatly encourages collaboration between different branches, which greatly enriches your work. And San Francisco is a special place; it's full of creative and innovative people, but it also has its challenges. The access to brilliant minds and resources is incredible, but it's an expensive place with many inequalities.

It's the best time to be in design; everything is yet to be done, and that makes it incredibly exciting

What is a Model Designer and what skills do they need?

A Model Designer works on designing AI models from an ethical and user-centered approach. It's not just about designing the interface, but providing context and feedback to the model. This involves closely collaborating with engineering and understanding how models are trained and how to incorporate feedback. Essential skills include knowledge in information architecture, prompt writing, taxonomy, and a great understanding of the user. It's also crucial to know how to work with data: ensuring it is clean and unbiased. Collaboration with engineering and Data Science is fundamental to ensure the model works as expected.

Our responsibility as designers is to ensure that what we design does not harm any human, neither psychologically nor physically.

Ethics in Model Design

We are in the prehistoric era of AI agents. Designing for them presents a huge paradigm shift.

How is ethics worked into model design?

It's a complex topic. What is ethical for one person may not be for another. At Google, we talk about Human Risk Engagement: minimizing the potential harm of interactions with AI. For example, designing models that do not discriminate and are safe for all people. However, ethics does not always easily translate into concrete actions, because it is a constant debate. That's why it's crucial to establish clear frameworks and tools that guide daily work.

We are in the prehistoric era of AI agents. In the future, we could have specialized agents collaborating with each other, eliminating the need for multiple applications. This will require rethinking how we design interfaces and processes. For example, I propose creating agent personas to better understand the needs and limits of the agents. These agents will be able to reason, plan, and work autonomously on certain tasks, but they should always be under human control.

Are we late to the AI revolution?

It's never too late. We are at an exciting moment where everything is yet to be defined. It's the best time to learn, experiment, and shape the future of design. The key is to stay curious and seek solid references.
This is the time to create, not to fear being left behind. If you're interested, now is the perfect time to get involved.

We are in the prehistoric era of AI agents. Designing for them presents a huge paradigm shift.

I suppose some of the concepts we've talked about in this interview might have blown your mind a bit. I don't know about you, but since I decided to dedicate myself to technology more than 20 years ago, I've always felt like I'm living in a unique and incredible moment in human history. But I must admit that what we've seen in generative AI over the last two years is truly overwhelming. If you're not following it with the attention it deserves, I recommend you keep an eye on the advances being made month by month, week by week, day by day, and especially use it because your work has already changed. Leave us a comment explaining what you thought of the episode, what surprised you, or if there's something you'd like to know more about. Don't forget to share it. And if you haven't subscribed yet, you know how algorithms are. Any like, heart, or interaction is always a great help. A big hug and I'll see you in the real world.