YouTube AI Recommendations
Redesigning YouTube's Smart TV UI for more purposeful relaxation.
Duration
2024 September–December
Course
Service Marketing & AI at Sungkyunkwan University
Services
#
UX Design
#
Prototype & Usability Testing
Challenge
A pre-ideation survey revealed that users come to YouTube primarily for entertainment, with finding something to watch as the top pain point. Smart TV users in particular tend to seek longer-form content — videos they can leave playing in the background — but feel unmotivated to use YouTube's existing discovery features like search or Explore tabs. Instead, thumbnails capture their attention and pull them into extended home page scrolling.
This pointed to two design questions:
How might we help users find a preferred video quickly?
How might we surface satisfying recommendations whether or not the user knows what they want?
Solution
We redesigned YouTube's smart TV interface with three new features targeting passive and active discovery:
Capped recommendation categories — Limiting the home screen to three categories reduces the cognitive load that leads to doomscrolling.
"Pick a video for me" — An AI-powered button within each category lets users choose a content mood without committing to a specific video, lowering decision fatigue.
Live Surfing — A channel-surfing mode built on YouTube's live content, simulating the lean-back experience of traditional TV while still reflecting the user's interests.
Testing & Findings
We built a prototype and ran seven usability tests across the three features. Users responded positively: they appreciated the reduced scrolling, felt that finding something new to watch took less time, and recognized Live Surfing as a nod to traditional TV. The feedback validated that both the active and passive discovery flows addressed the core pain points surfaced in our initial survey.