How do you know whether an AI response is actually good? This week Think First features three researchers at Microsoft AI who are center that question in their work. Christopher Monnier and Chuck Kwong, both Principal UX Researchers, and Wendy Wang Mertensmeyer, who leads the Copilot UX Research team. Together, they’re evaluating AI outputs with the user in mind, looking beyond model performance to understand what people actually value, what they are trying to accomplish and whether an AI experience delivers meaningful value in the real world.
Microsoft Design
Technology, Information and Internet
Redmond, Washington 14,465 followers
Stories from the thinkers and tinkerers at Microsoft. Sharing sketches, designs, and everything in between.
About us
Telling the stories and latest news around design and user research across Microsoft
- Website
-
https://microsoft.design/
External link for Microsoft Design
- Industry
- Technology, Information and Internet
- Company size
- 10,001+ employees
- Headquarters
- Redmond, Washington
Updates
-
What does civic participation look like in a design system? 🗳️ For a campaign with @TurboVote, the team built a visual language around the ballot itself using it as a visual device to make the vote feel tangible. Bold type, tactile textures and fragments of personal voices helped turn a practical message into something more energetic and human building a link between the act of voting and communities and change. https://lnkd.in/eTTX6NZr #NationalVoterRegistrationDay
-
-
-
-
-
+11
-
-
Microsoft Design reposted this
This November, we’re taking it to the ballot. Voting starts with being ready to vote. That’s why Microsoft is partnering with TurboVote to make it easier to register, learn what’s on your local ballot, and find the voting information you need. Your voice matters. Make sure you’re ready to use it. https://msft.it/6043abxib
-
Internal research from Tanya P. on metered AI use shows that when left to instinct, workers either overspend by defaulting to whatever's easiest or underuse AI by sticking to workflows they already trust.
Your AI bill has a new variable line. It's the usage-based cost that stacks up as agents take on more work. Many organizations are sending everything to the most advanced model available and not considering how that will impact the bill. As agents move deeper into daily work, that gets expensive fast. In my latest newsletter, I discuss the lever most organizations aren't pulling: the discipline of choosing the model that fits the required outcome. Choosing the right model, or combination of models, for each task turns AI spend from a one-and-done decision to a continuous management practice. It will be governed like budget, headcount, and vendor contracts already are. Learn more in this edition of AI @ Work:
-
The potential of AI-generated research wikis, telemetry analysis, role-specific research agents and personal AI chiefs of staff are among the many things changing how the industry is approached. Instead of using AI to summarise transcripts or speed up individual tasks, UX researchers are rebuilding their workflows around systems, context and agent-driven processes. Read thoughts from Ben Hanrahan, Christopher Gunderson, PhD, Sara Chizari and Bruce Phillips in the first edition of the new season of Think First.
-
What if typography could change the way you think? Research suggests it can. The way text is designed influences far more than aesthetics. It shapes credibility, attention, creativity, and even how we judge ideas. From the psychology behind readability to Microsoft's decades-long investment in type design, Rob McKaughan explores why seemingly small design choices can have an outsized impact on how information is perceived. https://lnkd.in/eSQwCzkW
-
-
When building agents, we often focus on accuracy, latency, and reasoning quality. But users start forming opinions long before they evaluate the output. They decide whether to trust, continue, or abandon an experience during the earliest moments. Drawing on user research with first-time users of a sales agent, Pooja Dhaka and Garima Sikdar outline five principles for designing onboarding experiences that build confidence and momentum from the start. These include delivering value before effort, making progress visible, reducing setup friction, and setting expectations early. Because every new capability, workflow or behavior change is another first impression. https://lnkd.in/eut4igpF
-
-
-
-
-
+3
-
-
When everyone can make something good, "good" stops being a differentiator. Designers can describe an interface and watch it appear. Build an app between meetings. Generate polished artifacts in minutes. Matthew S. explores what AI has really exposed: not our ability to make things, but our need to have something worth saying. Perhaps the competitive advantage that remains isn't production. It's conviction. https://lnkd.in/epjPvKVZ
-
-
Microsoft Design reposted this
Some of the most valuable design work disappears when the meeting ends. It lives in the designer who has seen a hundred versions of the same problem and can sense what needs to change. I wrote about what happened when we tried to make that judgment portable. It started with Kayla Brooks who turned Fluent Responsible AI guidance into a rubric. Her rubric became the foundation for a Responsible AI audit skill. The result cuts a 30-to-45-minute RAI review to six minutes while keeping evidence, uncertainty and human judgment visible. Full story in article: When Judgment becomes a material: https://lnkd.in/gckiSsK9 #ResponsibleAI #DesignSystems #ProductDesign #AIAgents #UXDesign Microsoft Design
-
What happens when someone who has spent decades supporting children in orphanages, refugee camps and conflict zones helps shape the future of AI? For Tanya P., UX research isn't just about understanding how people use technology. It's about understanding what people need—and building systems that help them thrive. In her essay, Tanya reflects on how running her nonprofit, Baal Dan Charities, has shaped the way she approaches research, accessibility and AI at Microsoft. Two worlds. One mission: helping people feel seen, supported and capable. https://lnkd.in/esTxfWiP
-
-
-
-
-
+2
-