UXLx 2024 — Wrap Up — Talks Day
24 May was the last day of UXLx 2024. And the last day means Talks Day. 🤩10 industry-leaders hit the stage to deliver their presentations to a crowd of 500 UXers from 38 countries, who were eager to learn and get their spirits refreshed with new ideas.
The presentations covered multiple UX hot topics — from AI, to design systems, leadership, accessibility, research, collaboration, and content. This wrap up gives people who were with us at the event the opportunity to revisit what they’ve learned. Those who didn’t have the change to attend can have a sneak-peek of what’s being discussed in the UX field lately.
🤝 UXLx is always a place of collaboration and that extends to this wrap-up article. Feel free to highlight, add any other takeaways we might have missed, and share what resonated the most with you.
24 MAY
🕘 MORNING
- AI by Design, Dan Saffer
- Guiding UX with Behavioural Signals and Settings, Lauren Alys Kelly
- Leading Successfully, Through Leading Ourselves, Aaron Irizarry
- Tough Talks and How to Have Them, Meghan Casey
- Turning Conflict into Collaboration, David Dylan Thomas
🕑 AFTERNOON
- Design System Lies, Stephen Hay
- Tell, Don’t Show: the Mass Misuse of Microcopy, Relly Annett-Baker
- Scaling your Scrappy Research Process, Danielle Green
- Innovating with Accessibility, Charlie Triplett
- Designing Products Powered by AI
AI by Design by Dan Saffer
With a wide and long experience within the Design field, Dan Saffer is a designer, author, and currently Assistant Professor of The Practice at the Human-Computer Interaction Institute at Carnegie Mellon University.
We kicked-off the last day of UXLx 2024 with Dan joining us to talk about one of the hottest topics of the moment — AI, and how it intersects with design.
One of the main thesis of the talk was that some of the greatest benefits of AI are from narrow AI, the one that focuses on mundane tasks. It’s a type of AI that’s high value, doesn’t require incredible perfomance from LLMs, and is available right now.
AI definition — AI is a digital system that makes an inference to address uncertainty.
Dan then poses the question — Do you actually need AI all the time? If an outcome is certain, we can simply use good design and the tools we’ve been using.
“Sometimes we don’t need a chatbot. We just need a better website.”
AI can be uncertain and unpredictable, which is uncomfortable for designers, PMs, etc., who’s job is to make everything certain and clear. When designing for AI, we also need to design for failure.
The AI sweet spot is where there’s not high risk, and it doesn’t require a lot of accuracy. This is where most companies fail.
Dan explained the concept of ‘AI innovation gap’. On a scale of buildable and desirable, data scientists want things technically challenging, not caring much if it’s desirable. On the other hand, designers want to build something desirable, but that might not exactly be buildable. However, the AI sweet spot is something very buildable and very desirable. This is why 90% of AI projects fail.
“Most people when they think about AI, they think of really hard things to do, and they overlook these easy but valuable things it could do.”
Researchers at Carnegie Mellon found out that the most successful AI projects combine high value (to users and the organisation), low risk and require modern technical performances.
Dan walked us through the interesting metaphor of Drunk Island, that’s basically making the question — “what would you give an island of drunk people to figure out?” We wouldn’t ask an AI to book us flights for a 6-week travel itinerary, but we might ask them what are some fun things to do when we’re abroad. We should think about this when we’re thinking about the type of projects we want to explore.
We need a combination of three methods, that together make a repeatable process that we can use to create AI:
- User-centered design — We need to understand the benefit to the user, and ask ourselves ‘Does the user need this?’ or ‘Would the user accept this?’;
- Service Design — Is there any kind of benefit to the organisation itself? Why spend money on an AI that will have no return?;
- Matchmaking — Connecting the capabilities of AI to human activities, takes a very basic capability and matches it with needs and activities to come up with a concept.
Designers who are most effective at working with AI have internalise a set of AI capabilities and examples. We need to learn the limits and capabilities of this new tech.
Three things AI is really bad at and humans are good at:
- Context — AI doesn’t not understand our users, our company;
- Taste — We have to apply our design training to determine whether something being produced by AI is actually good;
- Common sense — We have to determine what’s not going to work, what doesn’t make sense for users and for the company.
“The purpose of design isn’t just to solve problems. It’s to make the future more humane.”
Guiding UX with Behavioural Signals and Settings by Lauren Alys Kelly
Lauren Alys Kelly has been bridging the worlds of behaviour science and design for about 10 years. She’s the founder of behav, a Manchester based innovative behaviour thinking lab.
At UXLx, Lauren started by showing us see how we are so bad at changing our own behaviour. How many of us set ‘start exercising’ as a New Year’s Resolution? And how many of us fail to accomplish it and end up cancelling our gym membership the following month? We all try, and yet we all struggle. These bumps are everywhere in life, and also in the experiences we design. Prove of that is that retention drops, sessions are shorten and apps are abandoned. Behaviour can help us counter this if the experiences we design support our users in behaviour change. We must then start seeing experiences as pathways. Lauren guided us through the five steps of Behaviour Change Pathways:
- Start —The first step into a new behaviour, where things should be made as easy as possible;
- Stop — How do we stop behaviour? It’s about asking what are the triggers to the unwanted behaviour;
- Evolve — You can’t ask to much of people in the start, steps should be small;
- Realign —Identify where people are straying and correct them back to the right path, by providing soft opportunities to go back to the wanted behaviour;
- Modify — Adapt to the changes in a chaotic world; something to accelerate the behaviour, and avoiding it to stop; provide options to people.
“Change in behaviour change is not about triggering people; it’s about supporting people through behaviour change.“
Leading Successfully, Through Leading Ourselves by Aaron Irizarry
Aaron Irizarry is the Senior Director of Design at Capital One, and the co-author of “Discussing Design: Improving Communication and Collaboration Through Critique”. At UXLx, drawing on his extensive experience leading teams, Aaron joined the stage for a presentation on becoming a better leader, by learning how to lead yourself.
As a leader you’re responsible for nurturing your team, for making them feel good in the workspace so they can do great work.
“Let’s think about leadership as taking care of people, as taking care of one another.”
And how can we achieve that? ⬇
Self-awareness — Recognise and understand our moods, emotions, and drivers, and how that affects others;
Self-regulation — Developing a filter, think before acting. Develop the ability to control or re-direct disruptive impulses and moods.
Empathy — Empathy not only for our users but for the team we’re working with. Understand other people’s emotions, and treating them according to their emotional reactions.
Social skills — Learning how to build and manage relations. Build the skill to find common ground. Better than always being right, and not bending your opinion, is finding compromise.
Lead with vulnerability and transparency. If we’re not transparent with our teams, we erode trust. Encourage the team to be transparent. Show your team that they can talk to you. Being open and direct, but from a position of care, that encourages conversations.
In order to become effective leaders that create space for teams to be successful, we need a constant journey of adjustments. We need to adapt our communication strategies and leadership approaches.
Teams that are micromanaged perform badly. So we must lead through autonomy and enablement. We can do so by modelling an autonomous working style. We need to acknowledge that people might do things differently from what we would do. And that’s fine as long as there’re progressing to the outcome you’re intended, as long as there’s a share understanding of what you’re trying to accomplish. By wanting people to work the way we work, we’re limiting them from being themselves.
It’s important to check in regularly with our team. Ask them if them feel supported, if the way we’re working as a team is helpful, so we can know if there’s a need to adjust and adapt.
Establish personal norms — Before taking care of our team, we must take care of ourselves. Manage our time to not be stressed and exhausted, because that will unable us to be our best. We should invest in our own personal development. We should establish healthy routines so that we have the necessary energy and our emotions are regulated.
Vibe check — Our vibe sets a tone for the space in which our teams work.
“Encouragement matters. Saying kind words matters. Let’s be those kind of leaders.”
Tough Talks and How to Have Them by Meghan Casey
Owner of Do Better Content Consulting and author of “The Content Strategy Toolkit: Methods, Guidelines, and Templates for Getting Content Right”, a go-to guide for every content professional, Meghan Casey joined the final day of UXLx not to talk specifically about content, but about tough talks.
Meghan walked us through an approach to planning for and facilitating tough talks with stakeholders — the Daisy Framework:
- Define — What’s going on and why does it matter? We can leverage the Iceberg Model tool to help us think about the situation from both a micro and macro perspectives.
- Anticipate — What’s going on with them? We can use a quite familiar tool of UX professionals — the Empathy Map — to help us think about the situation from the perspective of the person we need to talk with.
- Introspect — What do I need to consider about how I might show up? For this, we can use the “What’s in My Head Check” tool, where we question ourselves how’s our ego, am I more concerned about the impact on me than the impact of what happened?
- Speak — Plan and have the talk. For this step we can apply the Meeting Mini Brief tool where we set the purpose, our & their mindset, key messages we want to communicate, desired outcome and the agenda, how will we approach the conversation. Additionally, we can use the Meeting Format Decision Matrix tool to decide whether the talk will be a face-to-face meeting, or a virtual one, depending on behaviour change/resistance and on the severity/priority.
- Yield — Pave the path forward. At this point, the Summary Sentence Starts tool can help us formulate the conversation summary.
Turning Conflict into Collaboration by David Dylan Thomas
David Dylan Thomas is the author of “Design for Cognitive Bias” and the founder of and CEO of David Dylan Thomas, LLC, where he offers workshops and presentations on inclusive design and the role of bias in making decisions.
David started by making us reflect on how both online and in-person we became really good at the SHOULD conversation. However, we’re not that good at the HOW conversation.
The Fundamental Attribution Error — For other we assume the behaviour has to do with their character, for us we consider the circumstances. (i.e. “Anybody who drives slower than you is an idiot. Anybody that drives faster than you is a crazy person.”)
Lately, the internet has become a place where you go find people who are doing it wrong and tell them they’re doing it wrong.
You get the conversation you design for. The colours and the objects present in the room influence people. The environment influences their behaviour. The same applies to internet design. While Medium’s colour and typography invites people to bring out their best, YouTube’s comment section seems that they don’t care of what you say and how. It’s like they’re saying “We put the least amount of effort possible into how this place looks so you should put the least amount of effort possible into how you conduct yourself”. Even the word choice influences people’s behaviour (i.e. Publish vs. Post).
Conversation Mindset — Collaboration vs. Hierarchy: Conversations based on hierarchy say “this is how it should be and there should be no deviation”. When you assume you already have the answer, you’re in a non-learning position. Collaboration says “here is a problem we can solve together”. This is a position of learning, that believes you’ll get better the more you learn.
There’s also a problem of language. We lack the language to talk about what we want. Most of the language is about how we don’t want. It’s should-oriented, rather than how-oriented.
Bureaucracy — a place where you try very hard to follow the rules.
Reactance — You can’t tell me what to do bias.
What’s the solution for this?
- Call out good behaviour
- Civil comments — platform to rate comments (no longer available)
- ReThink — AI-application that asks you if you really want to post something when it senses harmful content.
Question design — For any “should” question, understand what the goal of the proposed solution is, then frame a “how” question around that goal.
David walked us through this really interesting exercise called Eat Up, where you gather 8 people in a room to come up with ideas for a design question. This is an approach where you progressively combine the ideas of 8 people into one great idea. It has several advantages: mute the loud talkers; the DNA of everyone is on the final idea; and you get group diversity (from every department, from the CEO to the intern).
Does your business model want to generate winners or generate wisdom?
Three rules of productive discourse:
- Neither of us has the answer.
- Neither of us will win.
- We will listen to each other.
“Next time you see somebody doing it wrong, ask yourself if there’s a How conversation to be had.”
Design System Lies by Stephen Hay
With over 25 years of experience in digital design, Stephen Hay is the author of “Responsive Design Workflow” and the Creative Director at Rabobank, in the Netherlands.
Design systems are collections of design decisions. In turn, these inform new decisions. Stephen went through seven lies we tell ourselves when working with design systems:
Lie #1 — Most design systems are more than simply glorified component libraries. We focus too much on components. While they’re necessary, they’re not sufficient. Components become interesting when they interact within (and with) the environment. It’s important to seek understanding about the environment(s) around the design system.
Lie #2 — Documentation is optional. If you make a decision, it’s important to document it (who made the decision, why, what was the context). Documentation is the hub of your entire design system. If you don’t do it, people don’t know how to use your system.
Lie #3 —A design system should strive to be complete. The more tailor-made you make something, the harder it is to change. And the reason for a design system is that you can easily change things. Design systems should be designed for change, so you should keep things as simple as possible.
A design system is a process, not a product.
Lie #4 — “Atomic design” is a linear process. The most basic building blocks of a brand identity are the most important. Allow for the discovery using those, and introduce new patterns back into the system.
Lie #5 —A design system should predict the future. If you try to figure out what you’ll need, you might end up with patterns that people don’t need. Systems fail when you ignore the context so you can’t do tokens in isolation hoping the team will use them.
Lie #6 — Having choice replaces thought. The system is just a tool. You don’t want to only provide choices, you want to provide guidance to help people make the right decisions.
Lie #7 — Design systems are as important as we make them out to be. Design systems are not the end goal. The end goal is beauty. Aesthetics are a part of function, and form the purpose of the system.
Tell, Don’t Show: the Mass Misuse of Microcopy by Relly Annett-Baker
Joining us from England, Relly Annett-Baker is the Head of UX Content Strategy at Corporate Engineering, Google, where she runs content teams, writes content strategy docs, oversees content delivery, and she’s lost track of what else.
Relly promised a “fast and furious” tour, and she sure delivered it. She developed her talk around an analogy for enterprise websites and product with she story of her favourite house on the planet — the Winchester Mystery House in San Jose, California. A house that features doors and staircases to nowhere, windows without walls, which is hard to navigate around. This brought us to the question — How bad the UX industry still is at designing for complexity?
We struggle to think about the systems of users we’re serving and balancing their needs. Some of the best practices designers use can undermine any effort to negate necessary complexity. The established patterns of good design do not often make space for certain levels of complexity. Microcopy and product UI copy can eliminate deeper problems within the design and the experience. Relly walked us through a series of product anti-patterns:
№1 Vow of silence. The vow of silence is an interface that is “to cool to tell you what it does and how to do it.” Icons, illustrations, repeating videos, etc. could be minimalist back in the days when fewer people were using the Internet. Today not being helpful is not an option. Before you remove, try to refactor.
№2 Sticky bandage. A lot of products still face major discoverability problems, where people can’t find what to do. Too often they rely on solutions like product tours that highlight the apps amazing features, but that people simply skip and continue to not know what to do. Time and money put into that could be used to create a better interface, with clear in-line instructions, with a break-down of steps and progress.
“Don’t layer hints or directions on top or conceal them in tooltips. Anything that a user needs to complete the task should be visible on load.”
№3 Overeager assistant. We’re witnessing a widespread of unwanted chatbots that expect people to ask precise questions. Rather than implementing annoying chatbots, we should look for LLMs that support users. Interfaces should be adjusted, making space for more information for new users and less for regular users.
№4 The only customer is you. Different user groups deserve different experiences, adapted to their requirements. We should work towards interfaces where you have the space to contextually complexity. Not sweeping them under tooltips, for the sake of white space. However, reducing complexity is not the same as making it easy.
№5. To you this is a tool to get work done, to me this is Art. Design can involve art, but it’s essentially a craft, an intent to get to a particularly outcome.
“A designers primary job is to communicate information in a way that supports user intention and action.”
To create better user experiences, make the space in your interface for what users need to know, as they need to know it. Not in an help centre or a chatbot. We can harness AI to create better, more imaginative ways to support our users.
Scaling your Scrappy Research Process by Danielle Green
With nearly a decade of supporting organisations with UX Research and Product Strategy, Danielle Green is the co-founder of the UX Researchers’ Guild, and the Director of the User Experience Master’s Program at Claremont Graduate University.
Danielle invited everyone on the audience to reflect on the UX Research maturity of their teams, according to 4 different dimensions:
- Culture — The desire for research, the importance of research.
- Skill —The team’s ability to conduct rigorous research.
- Process — Efficiency, repeatability, and documentation of research.
- Impact — The scope of research and ability to affect change through research.
Those 4 dimension must then be evaluated according to 5 levels of maturity: absent, emerging, developing, maturing, and excellence.
This self-assessment will help us understand when it’s time to move from one research approach to another, and whether we need to improve our team’s UX Research culture, skill, process, or impact.
To sum things up, in order to change and take our research level to a new level we need to:
- Assess — The team’s research maturity;
- Identify — Agree on what need to change;
- Plan — Create a plan to improve research practices;
- Reflect — Define success and measure changes.
“Research change happens. One step at a time.”
Innovating with Accessibility by Charlie Triplett
Charlie Triplett is a champion of inclusive UX design and accessible UI engineering. He invented and open-sourced MagentaA11y.com, a tool for generating atomic level accessibility acceptance criteria, and wrote The Book on Accessibility is an operational guide for focusing any sized organisation on accessibility.
A common example used when we’re talking about accessibility are curb-cuts. Curb-cuts not only help people with a disability, they also help someone with pushing a baby stroller, someone older using a cane to get around, someone on crutches after surgery. A simple feature made for people with disabilities, made life better for everyone. You can include accessibility features into your designs, but like a light post placed in the way of a curb-cut, you can do it wrong.
Charlie addressed some accessibility myths in UX/UI:
MYTH. Accessible design is ugly. Several examples like the Apple store built for people with disabilities in Palo Alto proofs it wrong.
MYTH. Focus on the average user. Insights gained from designing and developing for people with disabilities, the people on “the edges”, are the best. The average people in “the middle” are only validating those insights. Disability is not just about a few people — 61 million adults in the US live with a disability. A number of others use and benefit from accessibility features in their digital devices.
Disabilities can be permanent and complete. However, we all experience temporary and situational disabilities.
MYTH. We have an automated tool that checks accessibility. A robot can take measurements, tell you if it meets the syntax or the code, but can’t understand if something is completely accessible.
In the beginning of the Web in mobile everything was too tiny, you had to pinch and zoom in, buttons were small. It was like having a vision and mobile disability. Then with responsive design (2010), we learned how to design mobile-first. This also made great things for accessibility. Digital curb-cuts with larger buttons (which helps people with motor disabilities), simplified processes, with less steps (which helps people with cognitive disabilities). Charlie claimed that since then not we’ve not experienced a big revolution, not much has changed.
“Average” for “average” yields sub-average results. When you design for “average” and try to apply to all, you end up excluding some, lose revenue and increase risk. It’s not where innovation happens. So how do we innovate? We need to find extreme use cases.
MYTH. We need some legal disclaimer on your website or have some special code injected. No legal disclaimer or special code can create innovation.
In the future, Charlie believes that we as designers and researchers can research and find ways to remove barriers people with disabilities experience with our products, design with inclusion, and end up with something we can apply to all.
Designing Products Powered by AI by Katrina Alcorn
Katrina Alcorn is a global product design executive, who recently led one of the world’s largest design practices at IBM, and the author of an award-winning book, “Maxed Out: American Moms on the Brink”.
To start, Katrina reflected on our changing relationship with machines. For millennia, our basic relationship with technology has been unchanged. It was a relation where we were the boss. Now, we’re entering a symbiotic relationship with machines, that’s becoming more of a partnership. Humans are becoming more willing to engage with conversational AI.
So how do you design for AI? Whether we’re designing for AI or not, as designers we must ask:
- What problem are we solving?
- Who are we solving it for?
- What will success look like?
Secondly, Katrina talked about setting a clear intent. You must align your product team to a clear intent. Think about what AI is really good at, and then think about what you’re trying to achieve.
“If you can’t map the purpose, the intent to one of this things, it might be time to step back and think whether you actually need to be leveraging AI.”
01. Efficiency: AI can automate routine tasks, from data entry to running complex analytics. This makes us more efficient and can also reduce the likelihood of errors.
02. Insights: AI can analyse large volumes of data to uncover patterns and insights that might be missed by humans.
03. Personalisation: AI can analyse customer data to offer a tailored experience to each user. Traditional computing cannot possibly achieve this at scale.
04. Preventing errors and downtime: AI can reduce costs and prevent errors and even accidents. This is particularly useful when it comes to managing expensive machinery.
05. Innovation: AI can super-charge innovation by working with humans to develop new products or improve existing ones.
Lastly, Katrina covered data and ethics. The quality, availability, and maturity of your data determines the value AI can deliver. When talking about data, we also need to address ethics. Some questions will come up: Where does the data come from? Do we have the right to use it? What could go wrong? Does it have bias?
“Ethics has to be something we’re all living and breathing. It’s definitely not something to be left to the end of the project to the regulators. It’s something that need to begin at the beginning of the project. And that means it starts with design. It starts with you.”
Ethics and design are inseparable. When you’re designing for AI, who you are, what you believe, and what your values are matters. Even if you consider yourself to be a very ethical person, it’s really easy to create a very unethical AI system if you’re not paying attention.
AI products are based on data and algorithms. The ethical character of these AI systems results from countless design choices made by us and our teams. Most of all, we have to care, that’s the foundation of all ethics.
Some final thoughts 💭
We might be living in a time where AI is overpowering the UX landscape. Where every day new AI capabilities are disclosed and products powered by AI infiltrate into the market. Some of us are responsible, or will eventually be responsible, for those products. No wonder the question ‘How can we design for (and with) AI?’ is now lingering our minds. We believe that Dan Saffer’s final statement really sums it all:
The purpose of design is to make to future more HUMANE.
We can be, and design experiences that are, more HUMANE by:
- Determining which AI capabilities really make sense for our users and the organisation;
- Supporting our users through behaviour change;
- Nurturing and encouraging our teams;
- Preparing for tough talks, where we care less for our ego and try to see things from the other’s perspective;
- Having conversations from a position of learning, based on collaboration;
- Creating systems that guide people, rather than restrict them;
- Guiding our users through complexity, visibly and as they need it (not by shoving off yet another chatbot);
- Self-assessing and helping our team take the next step towards better research;
- Removing barriers people with disabilities face when interacting with our products;
- Considering ethics and caring.
After Party — Sunset Cruise 🛥️🌅
After soaking in all the knowledge from the workshops and talks at UXLx 2024 it was time to partyyyy! 🎉 This year we booked an exclusive ship for a river cruise at sunset. 🌅
Food, drinks, music, scenic views of Lisbon, amazing sunset, great fellowship. Check, check, check ✔✔✔. All the ingredients reunited to end this edition of UXLx the right way. 🤩
See you next year? 😉
👀 Stay tuned!
We’ll start releasing the videos from the Talks Day on our UXLx Videos page, gradually, when we launch the next UXLx event. In the meantime you can check hours of content from the previous editions.
📸 All photo credits go out to UXLx’s official photographer — José Goulão.
