Barbie Bench: AGI Has Not Arrived
A content creator demonstrates Claude Opus's Barbie fashion designer 3D rendering project, highlighting both impressive and flawed outputs. While the model successfully created an interactive 3D game where users can dress Barbie and visit a fitting room, it struggled significantly with rendering accurate hands, facial features, and body proportions, leading the creator to conclude that AGI has not yet arrived.
Summary
The speaker discusses a project they assigned to Claude Opus 3.5 as part of a series of AI model capability tests. Unlike typical YouTube content featuring video game development with spaceships and conventional subjects, they requested the AI build a Barbie-themed 3D fashion designer game. The resulting application successfully implemented core functionality, allowing users to dress a Barbie character and access a virtual fitting room with 3D rendering capabilities. The creator acknowledges that Opus performed better on various aspects compared to other frontier models tested. However, the output demonstrates significant technical limitations. The 3D model has notably poor hand rendering, which the speaker emphasizes repeatedly. Additional issues include unsettling facial features described as 'terrifying,' awkward body proportions (oversized feet), and an awkward walking animation, though the character does have a 'sassy' walk. The speaker uses these flaws as evidence that despite impressive capabilities in some areas, current AI models have not achieved AGI-level performance, as they still struggle with anatomical accuracy and realistic human figure generation.
Key Insights
- The speaker tested Claude Opus by requesting it build a Barbie fashion designer 3D game with interactive fitting room functionality, representing a departure from typical AI benchmark tasks involving spaceships and conventional subjects
- Claude Opus outperformed other frontier models on certain aspects of the Barbie 3D rendering task, though the creator does not specify which particular elements exceeded competitor performance
- The AI-generated 3D model demonstrates critical failure in rendering human hands accurately, which the speaker identifies as a major technical limitation worth emphasizing multiple times
- Despite some successful implementation of features like character animation and interactive elements, the overall output quality—particularly facial features, hand anatomy, and body proportions—indicates that AGI has not yet been achieved
- The speaker characterizes the rendered Barbie's face as 'terrifying' and notes exaggerated proportions including oversized feet, suggesting current AI struggles with holistic anatomical accuracy in human figure generation
Topics
Transcript
[0:00] Everybody else on YouTube, all the bros are making um video or video games with like spaceships and all different stuff. Your girl wants a Barbie fashion video game. And so one of the Opus 55 um tasks I did was have it build Barbie um a 3D render of Barbie fashion designer. No model, no Frontier model has crushed this. You can basically like dress her and then go into the fitting room. This is the 3D the 3D model it made. Um, you know, I will say it did [0:35] some things better than other models. It like shaped its hair to her hands are Oh, agi is not not not arrived. Hands pretty terrible. That that…
Full transcript available for MurmurCast members
Sign Up to AccessMore from How I AI
I reviewed Opus 5.5 and GPT-6 Sol live - and the results surprised me
A live review comparing three newly released AI models—Opus 5.5, GPT-6 Soul, and GPT-6 Luna—where the reviewer conducts blind testing across multiple task categories and finds that while Opus 5.5 significantly improves on previous versions, OpenAI's models excel in creative tasks like SVG generation.
Warp’s Productivity dashboard is an eng manager’s dream
Warp's new productivity dashboard centralizes engineering team visibility for managers and CTOs, enabling real-time monitoring of development velocity, code quality improvements, and cost efficiency across distributed teams rather than relying on individual local setups.
She uses Claude Code to find a house with the right vibe
Hillary shares an AI workflow she uses for house hunting that leverages Claude Code to find listings and filter them against her subjective "vibe-specific criteria," which she describes as evaluating whether a space would feel good to wake up in and spend time at.
She hasn’t sent a birthday text herself in 6 months
A speaker shares how they've delegated birthday acknowledgments to AI for the past 6 months, having their AI assistant research friends' children's ages and purchase appropriate gifts on Amazon. They frame this as automating 'emotional labor' through AI technology.
JJ’s Claude Code skill sets up new projects for him
JJ discusses his favorite meta workflow for setting up new projects using Claude, which automates the creation of repos, folders, Claude MD files, and project structure. This skill significantly reduces setup time by leveraging trained knowledge on repository building practices.