Title: 对 AI 脸感到不适的，不只你一个人 | BestBlogs.dev
URL Source: https://www.bestblogs.dev/article/66a402f0a9?amp%3Butm_medium=feed&%3Butm_campaign=resources&%3Bentry=rss_article_item
Published Time: 2026-07-22 13:43:00
Markdown Content: 80
This article uses the Uncanny Valley theory to explain people's discomfort with AI-generated faces, and offers suggestions: accepting the discomfort, resisting aesthetic conformity, and focusing on content with a 'human touch.' 新 新周刊
Today 1009 words (about 5 min) View Source →
Sign in to highlight text and take notes as you read. Sign in now
中国青年报 2026-07-22 13:43 广东
AI脸受到厌恶，既应引发反思。
!Image 2 本文经授权转载自
中国青年报
（ID：zqbc yol ）
近期，“对AI脸感到不适”相关话题接连冲上社交平台热搜。
打开短视频平台刷微短剧，连续刷几部后，主角个个是高颅顶、大眼窄鼻、冷白皮，精致得挑不出错，可表情僵硬空洞，看着看着心里一阵发毛，关掉手机却说不清到底哪里不对。
!Image 5网友评论
这种说不清道不明的“不舒服”到底是什么？为什么会让人不适？
北京科学中心近日发文称，这种“似人非人”的违和感，便是“恐怖谷理论”。
> 1970年，机器人学家森政弘提出：当仿生物体的拟人程度持续提升时，人类对其的好感度会同步上升；但当拟人度接近真人，却又在某些关键动态特征（如眼神变化、面部肌肉联动）上出现可被感知的偏差时，人类的好感会骤然下跌，产生疏离、排斥甚至不适的负面情绪。这个好感急剧下坠的区段，就是“恐怖谷”。
（图/数字北京科学中心）
现阶段，算法虽能拟合出符合大众审美的静态五官，却难以复现人类视觉系统高度敏感的生物运动信号——例如瞳孔的细微缩放、表情肌的协同收缩，以及光影随动作的自然流动。这些动态线索的缺失或失真，会直接触发大脑的知觉冲突反应。
当人的视觉系统捕捉到高度接近真人的脸，却同时识别出多处不符合真实人类生理特征的细节时，便会产生排斥、不适的情绪反应与回避倾向。
（图/数字北京科学中心）
!Image 8 **对AI脸感到不适该怎么办？**
* **接纳不适感。**对AI脸感到别扭不是矫情，是人类知觉识别机制的正常反应，不必怀疑自己的感受。
* **拒绝审美趋同。**人类对面孔独特性的敏感具有进化心理学基础——面部差异帮助我们识别他人、判断意图与健康状态。不必被算法定义的“完美长相”裹挟，接纳真实的个体差异，本质上是回归人类正常的社会认知本能。
* **多关注“活人感”视频。**真正打动人心的内容，往往不只是靠无瑕疵的精致，带有真实细节的角色往往更有生命力。
网友评论
人民日报曾发表评论提醒，AI脸受到厌恶，既应引发反思，也应成为信号：可以利用技术压缩经费与制作周期，但不能用流水线产能替代艺术性和人文性；AI归根结底只是辅助性工具，专业的事还是要多交给专业的“人”；生产一百个“差不多”的产品只能争抢存量，创作一件“被记住”的作品才能创造增量。 真心换真心，观众愿意买单的，是被触动的感觉，它可能来自一句足够贴近生活的台词、一个会心一笑的动作，它的力量叫作真实。AI时代，有血有肉有“破绽”的内容，自有其珍贵。
总觉得AI脸怪怪的？你怎么看
排版：鱼尾
!Image 10   
读完点个【在看】👇
Key Quotes
> In 1970, robotics researcher Masahiro Mori proposed that as the anthropomorphism of a bionic object increases, human affection for it rises proportionally. However, when the degree of human likeness approaches that of a real person, yet exhibits perceptible deviations in certain key dynamic features (such as eye movement and facial muscle coordination), human affection suddenly plummets, generating feelings of alienation, aversion, or even discomfort. This steep drop in affection is the 'Uncanny Valley.'
> People's Daily once published a commentary reminding that the aversion to AI faces should not only prompt reflection but also serve as a signal: technology can be used to compress budgets and production cycles, but it cannot replace artistry and humanity with assembly-line efficiency. Ultimately, AI is just an auxiliary tool; professional matters should be left to professional 'people.' Producing a hundred 'good enough' products only competes for existing market share, while creating one 'memorable' work can generate incremental value.
> Accept the discomfort. Feeling uneasy about AI faces is not being overly sensitive; it is a normal response of the human perceptual identification mechanism. You do not need to doubt your own feelings.
Tags
AI Faces
Uncanny Valley
Human-Computer Interaction
Social Impact
Media Commentary
Related Articles
* Where Do Freshwater Fish Go When Swept into the Sea? Can They Survive? * Are Fourth- and Fifth-Tier Cities Racing to Host Concerts, and Will the "County-Town Middle Class" Still Pay? * Spending Over 1 Million RMB for a Hong Kong Bachelor's Degree * China's AI 'Four Little Dragons' Surpass One Trillion in Valuation has exceeded one trillion RMB, with the market pricing them based on 'definition rights' rather than traditional financial metrics, but the window of opportunity may only last 6-12 months.") * AutoNav GrowLoop: Building Rational Benchmarks for Empathetic Dialogue * Lychee-FD: Professor Zhang Min's Team at HIT Achieves Major Breakthrough in Full-Duplex Speech LLMs, Winning ACL 2026 Outstanding Paper Award * Is the Current Interface the Ideal Interface for Personal AGI? OpenAI President: Interfaces Will Disappear, Compute Will Never Be Enough! We Are on the Edge of the Agentic Era * After teaching a love course for five years, she discovered that the confusion of the post-00s generation has nothing to do with love * When AI challenges everyone's knowledge and abilities, what changes will we experience? | He Jiuheng, Yixi * OpenAI's Former CTO Brings a Prototype of an AI That Is Always 'Present' | Hao's Paper Talk, it analyzes the fundamental flaws of current AI interaction systems and elaborates on how the model achieves a truly 'present' next-generation human-computer interaction mode through 200ms micro-turn heartbeats, a unified multimodal architecture, and a dual-model design.")