Oct 22, 2024
6 min read
UX for AI Design: Key Insights from PromptUX Hamburg 2024
I’m back from PromptUX Hamburg, and it’s clear this wasn’t just another tech conference. Between the practical insights and actionable strategies shared, these few days have been insightful about the intersection of AI and Design.
I’m back from PromptUX Hamburg, and it’s clear this wasn’t just another tech conference. Between the practical insights and actionable strategies shared, these few days have been insightful about the intersection of AI and Design.
Press enter or click to view image in full size
PromptUX — Auditorium 1 before opening Keynote
The conference spread across 2 of CinemaxX’s main theaters, with the two main auditoriums hosting keynotes. The cinema’s seating meant a perfect sight for everyone — a welcome change from straining to see slides in typical conference halls. During breaks, the lobby became a bustling networking space, with discussions forming around the concession areas normally reserved for popcorn and movie snacks.
Sofie Hvitved: Redefining Personalization
Sofie Hvitved’s session on AI-driven experiences cut through the usual personalization hype. Her presentation explored the implications of these changes and what they mean for content creators, consumers, and society at large. In an era of rapid technological advancement, the landscape of content creation and consumption is undergoing a seismic shift. As we move from traditional strategies to more explorative and creative approaches, we find ourselves at the intersection of relevance, trust, and hyper-personalization.
Press enter or click to view image in full size
Sophie Hvitved
As AI technology advances, we’re likely to see a proliferation of average content, with some AI-generated content potentially competing with high-quality human-created content across various mediums, including newsletters, social media, commercials, TV, and radio.
From Many Channels to Hyper-Personalized Content
The content creation paradigm is shifting:
Today: Create one piece of content, and distribute to many channels.
Future: Create endless variations, and distribute personalized content.
This shift towards hyper-personalization raises questions about relevance and engagement in an age of “liquid content.” In the digital age, consumers are increasingly impatient and disloyal. The media industry must find new ways to create engagement. Liquid consumers demand tailor-made, seamless solutions, leading to hyper-personalized and curated experiences based on real-time user data and insights.
The Ethical and Societal Implications
As we embrace hyper-personalized content and curation, we must grapple with significant ethical and societal questions. How will this challenge the way we perceive content and reality in the future? The implications are far-reaching and potentially transformative.
Experts estimate that up to 90% of content could be synthetically generated by 2026, according to the Europol innovation lab. However, it’s crucial to acknowledge the inherent uncertainty in such predictions. The future of the internet is likely to be more complex and nuanced than we can currently imagine.
The key insight? Personalization isn’t just about preferences anymore — it’s about understanding and responding to user states in real time.
As we navigate this new landscape, we must ask: Does trust even matter in the future? We’ve been obsessed with trust, but perhaps it’s the wrong focus. In a world of hyper-personalized, AI-generated content, relevance might become the new currency.
The future of content is liquid, personalized, and AI-driven. As we stand on the brink of this new era, we must be prepared to rethink our approaches to creation, distribution, and consumption. The challenges are significant, but so are the opportunities. By embracing strategic foresight and adaptability, we can navigate this changing landscape and shape a future that harnesses the power of technology while preserving the essence of human creativity and connection.
Vitaly Friedman’s Deep Dive into AI Design Patterns
Vitaly’s session was dense with practical insights. His concept of “Intent-Aware Navigation” stood out as immediately applicable. Rather than theoretical frameworks, he showed working prototypes where navigation adapted to user behavior patterns.
Press enter or click to view image in full size
Vitally Friedman
Today, when we think of AI, we often think of a text prompt. A single text box, which we populate with context, keywords, pointers, guidelines, modes, and references. However, a text box isn’t an effective design pattern for complex input. How do we design better AI UX?
What patterns can we use to drive people towards their goals, faster?
Let users act directly on the output (more evidence, more insights, outliers)
Navigation elements that reorganize based on predicted user goals.
Let users zoom in and out of the content
Cherry-picking the best AI summaries to create their own
Allow users to be critical of the AI-generated summaries by providing the sources used by the LLM.
The value was in the details — every pattern came with implementation notes and details to watch for. I’ll be sharing these with my team as potential starting points for our upcoming work.
Konrad Piercey on Building Trust in AI/ML products
Konrad’s session on transparency in AI systems hit on a critical point: user trust isn’t a have, it’s essential for adoption. He shared specific examples of projects that succeeded or failed based on how well they communicated AI’s role to users. His framework for transparent AI implementation is something I’ll be using in my next design review.
Press enter or click to view image in full size
Konrad Piercey
He shared several innovative approaches to making AI operations visible to users:
Activity Dashboards
Personal AI interaction histories showing when and how AI influenced decisions
Clear visualization of data sources used for personalization
Timeline views of AI-driven changes to user experience
Real-World Example
He demonstrated these principles through a case study of a content personalization system where users could:
View their content consumption patterns
Understand why certain content was recommended
Adjust the weights of different recommendation factors
Reset the system’s understanding of their preferences
Export their interaction data
The results were compelling: implementations using this framework saw a 47% increase in user engagement with AI features and a 65% reduction in feature opt-outs compared to traditional “black box” approaches.
A chat about the real-world implementation of AI solutions with Josephine Scholtes and Mara Pometti
The joint discussion with Josephine Scholtes and Mara Pometti was refreshingly practical. They walked us through their company’s journey of implementing AI in design workflows and projects, including:
Initial proof-of-concept phases
Team structure adjustments
Technical hurdles and solutions
Cost considerations and ROI metrics
Wrapping up
The conference reinforced that AI in design isn’t about replacing human decision-making — it’s about enhancing it. The tools and patterns shared weren’t just interesting ideas; they were practical solutions to real problems we face daily in design work.
What’s your experience with implementing AI in design workflows? Share your successes and challenges in the comments.
#PromptUX #AIDesign #UXDesign #DesignStrategy #Pagerduty
Originally published in Medium: https://medium.com/prototypr/ux-for-ai-design-key-insights-from-promptux-hamburg-2024-0f3392f8f6d8
