The end of UX? This is the start of the service design of AI

If you’re a UX or product designer, there’s reason to be worried about job security. GenAI enables anyone to design interfaces using prompts and Figma is collecting data in a way that is threatening designers. Maybe it’s the end of UX, but more likely we’re heading to a shift towards needing service design more than ever.

The last fifteen years was a period that transformed design’s role in tech. Designers were able to accomplish more and take on increasingly complex tasks, in less time. Sketch, Figma, Zeplin, Framer, and the dozens of other new software changed what was possible.

But today the concerns about AI are different because it could replace design tasks. In the past, when technology significantly advanced designers adapted by taking one of two paths:

1. Design better solutions by leveraging new tools and knowledge (e.g. Create more efficient and user friendly apps); or

1934 chart from Raymond Loewy demonstrating how solutions became more complex as technology evolved and designers had a bigger toolkit

2. Design new capabilities by imagining new systems (e.g. Create platforms and services so that mobile apps are more impactful)

Example of how new systems enabled entirely new designs to be conceptualized

While front-end innovations get most of the attention —e.g. TikTok's addictive interface— the transformative disruption happens when your mobile device makes it easier to interact with the world around you.

Similarly, the mobile device in your hand was made possible by thousands of backend innovations that have happened over the last half-decade. Historically designers have had little influence into these backend systems and engineers controlled the capabilities you could work with.

Designers are feeling like AI will displace them by automating many workflows

For designers we have entered the uncanny valley, where software feels threatening. The concept was introduced by Masahiro Mori in 1970 to describe the human-robot relationship if and when they the machine look and act like us.

And for the thirty years of the web, designers had to struggle to make software do basic things. It took Adobe's Photoshop until 1998 to release multiple undos and until 2017 to support variable fonts.

Today's software isn't only assistive, it automates. Design systems have been hugely assistive to building products and scaling them across large deployments. But as Figma so directly put it in this blog post, the end goal of design systems is to automate. And GenAI will unlock this capability by automating many tasks and enabling more dynamic and adaptive UIs.

But GenAI makes a new material available to designers

The reason LLMs will be so transformative to businesses and to product teams is because they simplify the processing and analysis of large data sets. Orgs can now outsource their massive computing tasks to the LLMs, like ChatGPT, Gemini, Claude, etc.

While the current generation of LLMs excel at quantifiable and verifiable tasks —data/doc analysis, coding, comparing— but by 2030 GenAI is expected to achieve human-level performance on many creative tasks.

While the expectation that GenAI will get better at creative tasks should rightfully intimidate designers, huge leaps forward in making data more useful will change how products are designed. This futuristic capability of crafting meaningful interactions from data is an entirely new material for designers to work with.

Design leader Emily Campbell was our guest on the Design of AI podcast where she spoke in depth about how GenAI is reshaping design (listen to the episode).

The perspective is that GenAI will enable the design of interactions which are more situationally-aware. Just like when you type a search into Google, it infers your intent based on your keywords + search history + websites you have visited —LLMs will be able to add context to your interactions within applications. LLMs do this by referencing the wide data-set they are trained on, plus any additional fine-tuning or retrieval augmentation process (if available).

Evaluating and improving the situational awareness of LLMs is one of the great ambitions of AI researchers. And while the technology's capabilities is still improving, designers today can get shockingly good results by leveraging fine-tuned LLMs in narrowly-focused tasks (e.g. interpreting content, summarizing docs, enhancing search, evaluating against a benchmark).

This new design material can simplify and eliminate unnecessary steps from cumbersome user interactions:

• Acting as an assistant seeking out content, assets, insights

• Delivering a first draft based of what you need to get done

• Turning a join flow into a highly personalized set up process

• Predicting and processing a job to be done based on an example

• Consolidating docs and data into summarized and actionable insights

• Generating potential new creative directions for an asset based on guidelines

The list above are what GenAI can do today and designers should be considering these now.

Understanding GenAI and how to design with it

We've all had to choke on the insane claims about how AI will change the world around us. But so few people actually understand what AI is today and the best ways to leverage it. The problem is compounded by investors and founders laying out extremely ambitions visions for how to leverage AI that often can't be accomplished today. That's ultimately why Brittany Hobbs & myself started Design of AI to bring in experts to talk about the reality of AI.

Jess Holbrook is the Head of UXR at Microsoft AI and has been working at the forefront of integrating academic insights into commercializable products. In a recent Design of AI episode he focused on product teams needing to understand the tech and also to seek clarity from leadership about how "throwing AI on it" actually works within the business model and how it enhances the lives of users (listen to the episode).

But as designers learn to use this new material, they must consider:

• GenAI is very effective at specific tasks and doesn't add significant value in others

• GenAI is experimental, requiring teams to learn by failing and users to be trained & guided

• GenAI isn't about replacing interfaces, rather about reducing the effort to accomplish tasks

• GenAI isn't going to grow your user base and revenue without a comprehensible, value-driving, and ethical strategy that aligns to your business vision

Most importantly, successfully integrating the above requires designers to move from creating interfaces and flows, to instead obsess with how to improve services holistically.

Why service design is more important than ever

As a discipline, service design isn’t new. The Service Design Network has been operating globally since 2004 and most corporations in regulated industries have hired service designers.

Service designers are amazingly skilled at facilitating and planning change. UX designers are amazingly skilled at creating change and transforming experiences. You'd imagine that they would collaborate exceptionally well. Last year Daniel Tuitt wrote a great article making a case for why UX designers should apply service design principles/ thinking.

But service design has failed to gain significant traction.

They're hired to re-imagine customer journeys —to make various corporate groups operate more effectively from the customer's perspective— and deliver research, plans, documentation to plan the necessary changes. The problem is that teams are structured to be siloed, each with their own responsibilities, targets, and remits. There are very few incentives to prioritize working on cross-organizational initiatives rather than immediate quick fix projects that help you get promoted.

As a strategist who applies service design methods at PH1 Research, I can't believe how unreceptive orgs can be to learning about systemic gaps and opportunities.

But GenAI changes all of this. Organizations are rushing to leverage the new technology, inspired by growing their revenue, plus the fear of the looming threat of being disrupted.

Product and engineering teams are being challenged to deliver significant gains and AI's best opportunities are in examining potential improvements across the "front stage" and "back stage" —terms borrowed from service design to describe the holistic process of delivering value to customers.

Interaction Design Foundation: Frontstage and backstage are the areas that border the line of interaction in a customer experience. Customers directly encounter frontstage parts, such as counter staff, but not backstage ones, including back-end staff, systems and other partners. In the best experiences, frontstage and backstage operate in harmony.

Why service design is more important than ever

However, the roll-out of GenAI isn't going as well as leadership had hoped. A March 2024 report by MIT Technology Review Insights (MITTR) found that orgs are very bullish on the tech but had underestimated the effort and resources required to implement generative AI successfully.

“A wide pool across all industries continues to experiment with these powerful tools which can deliver practical benefits to certain use cases,” Williams adds. “To unlock the transformative capabilities that will separate the leaders from the followers across any given sector, executives will need to start building end-to-end capabilities that allow them to handle large datasets, accurately contextualise the data for business value and ensure the responsible and ethical application of AI.”

Achieving this level of transformative, cross-organization change necessitates the unique ability of designers to look at challenges from different perspectives. While UX designers have excelled at the granular digital experience, service designers have taken a zoomed out perspective, evaluating from the perspective of the business and necessary services.

Standard service designer, Morgan Miller, detailed this out and shows the matrix of possibilities for designers who want to expand beyond the pixel and design system.

Both service & UX designers are well-equipped to benefit from the rush to create value by implementing GenAI. But UX designers have an advantage in this rapdily-changing job market because they have a bias to action.

The obvious path forward for UX designers in AI-hungry orgs would be learn the traditional deliverables of service design —journey maps, storyboards, service blueprints, service prototypes— but the greatest benefit of the methodology is facilitating cultural transformation.

The art of facilitating change processes will be more essential than the service design deliverables because successfully implementing GenAI will require orgs to change how they think about delivering value.

Creating change requires assessing the overt and covert factors impacting the system

The path forward for UX designers looking to adapt to the service design of AI

While this will be time of significant change, its one that offers an opportunity to create experiences that significantly impacts customers and the business. And its part of the natural evolution of design as technology evolves.

GenAI is a new design material that will enable you to simplify user journeys and help solve irritating problems that have been frustrating you for years. You'll have the ability to influence the backend of UX and turn mountains of data into useful and actionable outcomes.

But to take advantage of this, UX needs to learn from the fabulous work of the service designers who have facilitated and planned change within many of the world's largest private and public organizations.

UX designers who want to take on the design of these complex systems will need to incorporate core pillars of service design:

• Design holistic solutions that align the backstage and frontstage around shared outcomes

• Zoom in and out to evaluate from the perspective of the business, stakeholders, users

• Learn the art of facilitating change within cultures

If you'd like help with this process, please reach out to me, or book a meeting with PH1 to work on your AI project.

For in-depth AI knowledge, follow the Design of AI community and podcast: Spotify, Apple podcasts, Youtube.