Contra Labs盲测:真人设计师在76%的对比中胜过AI,Fable 5.2表现最好
真实客户简报加盲评,比常见的自评榜单可信多了。人类设计师还赢着76%,但Fable 5.2和Kimi K3.1追得比想象中近。做设计或选型的人可以看看哪些模型已经接近能用的水平。
Contra Labs组织了一场AI设计能力盲测,邀请专业设计师基于真实客户需求进行盲评。结果显示人类设计师在76%的对比中仍然获胜,其中Fable 5.2是表现最好的AI模型,Kimi K3.1也获得了较高评价。
测试方法采用盲评形式:设计师无法得知作品出自人类还是AI,凭真实项目简报的质量判断优劣。这一测试避免了主观预设,结果反映的是实际设计输出水平的直接对比。Aran Komatsuzaki评论称,AI与人类的差距比预期小得多,Kimi K3.1的表现令人印象深刻。
该结果对设计行业有参考意义:AI生成设计已在真实客户场景中接近可用,但专业设计师的整体优势仍然明显。对于评估AI设计工具的团队来说,基于真实简报的盲测比传统基准更能反映实际能力。
Interesting result from Contra Labs on AI design. Pro designers voted blind on real client briefs, and humans still win 76% of comparisons, with Fable 5.2 the best model.
Honestly, the margin is much smaller than expected, and Kimi K3.1 seems great.
Giveaway for your answer to how you can tell a design was made by AI.
This report is categorized under AITOP's "Models" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include Fable 5.2, Kimi K3.1, Contra Labs, design.
Source attribution: curated from Aran Komatsuzaki (论文推介) (@arankomatsuzaki). Cross-source corroboration: 6 independent sources have covered this development (including Epoch AI, kimmonismus, AI Will), confirming industry-wide signal strength.