Research in the Automated Future
Summary
Ovetta will talk with us about reinvigorating the practice by incorporating Design Anthropology into our research tool-kits and further broadening our set of methodologies to include new research methods for AI/ML design.
Key Insights
-
•
Design and research are tightly intertwined; it's often impossible to tell where one ends and the other begins.
-
•
Machine learning models primarily predict based on past data rather than interact dynamically with the environment.
-
•
Intelligent systems involve multi-agency contexts where both humans and machines have decision-making power.
-
•
Unlike static software, AI systems are dynamic, learning and evolving based on interactions and data.
-
•
Design research for AI must be future-focused and speculative, addressing not just usability but cultural values and trust.
-
•
Bias in AI is complex, involving mathematical, statistical, cognitive, and human biases requiring nuanced mitigation.
-
•
Researchers and designers need high AI and data literacy to influence data inputs and model fairness, not just outputs.
-
•
Explainability in AI is challenging, especially with neural networks, necessitating governance and 'glass box' approaches.
-
•
Design anthropology and participatory methods help uncover unspoken human practices important for automated systems.
-
•
Speculative prototypes (provocation prototypes) and perpetual synthesis support ongoing learning about future AI use.
Notable Quotes
"Design is the conscious and intuitive effort to impose meaningful order to chaos."
"Our role as designers and researchers will be to determine what not to design to preserve human culture and values."
"Machine learning models can see, hear, think, and act but are limited to problem-solving based on past observations."
"The relationship between a user and AI system is multi-agency; both have agency to act and influence outcomes."
"The nuance is the enemy of machine learning and AI; they struggle with ambiguity where humans thrive."
"Research for AI must go beyond single agency frameworks to consider complex ecosystems and multi-modal environments."
"It's not about looking under the hood of the model, but about what data goes into the hood that you can influence."
"There are many types of bias—mathematical, statistical, cognitive, human—and they affect AI outcomes differently."
"Explainability requires transparency at every stage with governance ensuring fairness and equity in model building."
"Speculative research requires participatory methods, putting people into future scenarios, because people often can't articulate what they will want."
Or choose a question:
More Videos
"Reflect on what you experienced internally and externally to sharpen and reveal new insights."
Deanna MitchellDesigning with culture: Unlocking impactful insights for Product and UX
March 12, 2025
"Behavior over time is your culture; it’s how you choose to behave and incentivize others to behave."
Adam CutlerPeople + Places + Practices = Outcomes
June 8, 2016
"When blind users choose between Android and iPhone, they weigh trade-offs between stability and customizability."
Sam ProulxUnderstanding Screen Readers on Mobile: How And Why to Learn from Native Users
June 6, 2023
"If there’s a community that’s already lesser heard, then it’s even harder and those voices are often drowned out."
Shazia Ali Bruce Gillespie Joyce Lee Andy WarrCommunication: Innovative techniques for making your voice heard [Advancing Research Community Workshop Series]
August 21, 2024
"I simply cannot wait for you all to see the outstanding presentations they prepared."
Bria AlexanderTheme Two Intro
September 8, 2022
"Design methods we love, like sticky notes, aren’t accessible to everyone and can exclude some people."
Saara Kamppari-MillerInclusive Design is DesignOps
September 29, 2021
"Contextual navigation is really powerful — these AI-generated prompts help you dig deeper into a topic."
Louis Rosenfeld Peter Van DijckGenAI for UXers: A Rosenbot Demo and Discussion
September 11, 2025
"Our role as designers and researchers will be to determine what not to design to preserve human culture and values."
Ovetta SampsonResearch in the Automated Future
March 11, 2022
"We use a participant matrix and assign pseudonyms to protect personally identifiable information throughout research."
Lisa Spitz Nikki BrandBuilding Trust Through Equitable Research Practices
November 18, 2022
