This video is featured in the Designing with AI 2025 playlist.
Summary
We built a novel math tutoring app for 11-year-olds in the UK. Since this was our first AI project, we expected lots of technical issues. Those happened and on top of it, we ended up questioning the value of designers in AI-driven products.
Key Insights
-
•
Designing AI products demands a strong understanding of AI’s inner workings to anticipate challenges and create meaningful alternatives.
-
•
AI model updates can unpredictably impact performance, requiring flexible design solutions and rapid iteration.
-
•
The Boost 11 Plus app used photography and step-by-step AI tutoring to assist stressed students and their families effectively.
-
•
Strict SLAs, like 24-hour design comment turnarounds, help teams manage rapid changes and keep alignment in AI projects.
-
•
Design principles such as avoiding gamification and clarifying the app is a calculator, not a person, were crucial to maintaining ethical clarity.
-
•
Designers face loss of traditional precision and predictability when designing with large language models due to their inherent opacity.
-
•
Rapid AI advances demand constant adaptation; designers must create multiple design alternatives to navigate ambiguity.
-
•
The plastics analogy illustrates how shiny new technologies can deliver powerful capabilities but carry unforeseen long-term harms and ethical risks.
-
•
AI’s environmental cost, like energy use for generating content, is significant and often overlooked in design and deployment choices.
-
•
AI systems can diminish critical cognitive skills like memory and critical thinking, highlighting the need for mindful usage and design.
Notable Quotes
"The app reads the question and prepares tutoring material, then helps solve it step by step or through chat if further questions remain."
"We realized designing for AI is very different from other computational mediums we used before."
"Understanding the lifecycle of AI code is essential because AI is essentially code running in constant evolution."
"With LLMs, precision is not possible; no one really knows why it decided what to say."
"AI is like plastics on the surface — it gives amazing powers but has hidden damages we must not repeat."
"Design is an excellent way to navigate chaos, turning ambiguity into direction through alternatives."
"We used design principles like look but don't copy, this is a calculator not a person, and no gamification to guide tough decisions."
"AI companies focus on the shiniest thing that comes without caring about stability or reliability."
"It costs about as much as a bottle of water in energy to generate a single email from an LLM system."
"AI users have reduced critical thinking skills within just two years of using AI-assisted tools."
Or choose a question:
More Videos
"Captions and subtitles were initially designed for people who are deaf or hard of hearing but over 50% of Americans watch content with subtitles."
Kavana RameshMeaningful inclusion: Practicing accessibility research with confidence
September 24, 2024
"AI makes data more accessible to everyone but won’t replace the need for validated, empathetic UX research."
Rima Campbell Amrit S BhachuIncrease Productivity and Drive Business Impact
September 24, 2024
"Please stand up and hold each other’s hands, and on three, raise them high and shout Yeah so virtual can hear you."
Bria Alexander Louis RosenfeldOpening Remarks Day 1
March 25, 2024
"We try to limit routine global calls and break large meetings into three regional meetings so people don’t have to join at impossible hours."
Kim LenoxLeading Distributed Global Teams
May 20, 2019
"Peter van D**e really pushed us to think about the bot as part of a conversation, not just a query-response machine."
Louis RosenfeldThe Rosenbot and the Rosenverse: An AMA with Lou Rosenfeld
June 5, 2024
"People in organizations are already, to some extent, out there talking to users."
Steve PortigalLooking Back…to Look Ahead
March 26, 2024
"We created a self-service UXR hub to empower teammates with research fundamentals amid downsized research teams."
Maggie DieringerCreating Consistency Through Constant Change
January 8, 2024
"We decided to take a broad definition of research repositories to lessen shame and embrace the complexity of current practices."
Brigette Metzler Dana ChrisfieldResearch Repositories: A global project by the ResearchOps Community
August 27, 2020
"Consistency between control labels and prototype instructions matters a lot, because mismatches can cause big problems for alternative navigation."
Sam ProulxPrototype Reviews, People With Disabilities, and You
October 1, 2021
