O conteúdo está em inglês; a navegação e os filtros mantêm o idioma selecionado.

The Generative UI Paradox: More Important, Yet More Disposable

As AI begins generating interfaces and frontend code on demand, UI may become the primary way to present model results. Its role in guiding decisions and actions could grow, while individual screens may become short-lived artifacts rather than assets requiring extensive polish.

Publicado em 06/08/2026, 13:01:34Última atualização 06/08/2026, 13:01:34Publicação editorial

The argument presents a central paradox of generative UI: model output may increasingly arrive as an interactive visual experience instead of plain text. Interfaces can organize data, choices, and actions, helping users understand results and respond to them more quickly. In that sense, UI could become more functionally important.

At the same time, on-demand generation may make many screens disposable. If a page exists only for a single task, minor differences in color, typography, or placement may not prevent users from reaching their goal. Traditional standards of visual refinement and consistency may therefore matter less for some temporary experiences.

Current practice supports AI’s ability to accelerate prototyping, marketing assets, and parts of frontend implementation, but generative UI still faces constraints involving complex requirements, design systems, brand coherence, delivery, maintenance, and real-world user needs. A more defensible conclusion is that AI may reduce routine interface production without eliminating the importance of usability, judgment, and long-term experience design.

Evidências das fontes 6

A verificação de fontes não garante exatidão permanente. Considere a fonte original, a data e as correções posteriores.

瑟瑟发抖，UI设计也要被AI支配了么？_酸梅干超人-站酷ZCOOLm.zcool.com.cn · Complementar

所以光生成一个界面是没有意义的，这个界面做的再好看再美观，也不代表它有真实落地的可能，只是完成了设计工作的一小部分而已。所以看到这里是不是感觉很熟悉？和我们在追波上看到的各类飞机稿案例没有太大的区别。第三点，上面的模型只是一个非常简化的流程，有过真实的项目经验就应该知道流程中会有反复拉锯的事件，需求的变动，设计稿的修改，方案的调整，这些零碎的工作去和 AI 提，你会发现有输入问题的时间，可能设计早就修改完了。尤其在最终的标注和切图环节，涉及到设计时对设计稿的制定，标注和切图实际上有很多主观性和经验化、场景化的输出要求，它需要设计师在经历整个项目流程后自行判断。除了以上三点，还有很多其它的问题难以解决，而这些问题的总和，决定了没有出现真正的人工智能之前，使用 AI 来输出 UI 界面是一件不靠谱的事情。只有视觉领域没有前期那么多条条框框，也没有后续一系列交付和维护需求的场景，才是 AI 真正产生价值和影响的方向。用 AI 做 UI 设计不靠谱，但是不代表对UI设计师没价值。这话听着可能挺拗口的，但事实如此，因为在国内的UI设计环境中，UI 设计师的工作内容往往不局限于产品设计一个领域内。还包含下面这些操作：我相信大多数 UI 设计师做平面和运营设计的感受就和上坟的状态是一样的，上坟起码是怀有对先人献花，做平面和运营设计只会产生对自己深深的唾弃…… 所以 AI 的出现可以很好的处理这部分的需求，比如一些虽然用处不大，但就是甲方还是老板就是想用的 3D 风格图标。或者，在登陆页用的比较滥俗的 3D 场景背景，为了这样的东西花很长的时间去建模和渲染，投入和收益是完全不成正比的。所幸使用 AI 也可以帮助我们非常短的时间内获取想要的效果。 [...] 还有就是各类运营设计图，互联网运营设计和一般的品牌广告视觉比较起来，就是对创意性的忽视，画面并不需要埋伏大量的隐喻、内涵或故事，就是单纯的应景和吸引用户注意力，这恰恰也是 AI 最适合输出的东西。包括广告 Banner、H5、启动页等，都可以通过 AI 快速生成一些高质量的插画，来替代我们自己从头开始绘制的苦恼。所以你看，对于我们职业价值不高的这些杂活，AI 都可以很好的去完成，而且可以肯定未来可以完成得越来越好，我们就可以节省更多的时间，不管是投入精力给体验和交互，还是准点下班，都符合