A review of research on universal usability, plasticity of user interface design and facilitation of interface development with universal usability is presented. The survey was based on 165 research papers spanning over fifty-five years. The foundations of adaptive or intelligent user interfaces (AUI or IUI) are presented, three core domains being focused upon: Artificial Intelligence (AI), User Modelling (UM) and Human–Computer Interaction (HCI). For comparison of the various AUIs, a proposed taxonomy is given. One conclusion is that an efficient training vector for fast optimal convergence of the machine-learning algorithm is a necessity, but key to this is the bounding of the dataset, the goal being to achieve an accurate user preference model, which has to be built from a limited number of datasets obtained from the human interaction. More research also needs to be conducted to ascertain the usefulness and effectiveness of IUIs compared against AUIs.

With the global mobility of users, interface design must take account of the abilities and cultures of users, derived from actual user behaviour and not on their feedback. A key question is whether the interface should be adaptive under system control or be made adaptable under user control. A need is identified for an “afferential component” that stores a priori information about the end user, an “inferential component” that determines to what extent the user interface actually needs to be adapted, and the “efferential component” that actually determines how the adaptivity is applied seamlessly to the system. Application to e-learning is a priority: the use of machine intelligence to achieve appropriate learnability, ideally enhanced by “Playful interaction”, was found to be desirable. Universal application of adaptation lies in the future, but AUI properties cannot be ascertained while disregarding the other parameters of the system in which it will be used. A more complete understanding of the human mental model is necessary, requiring a highly multidisciplinary approach and cooperation between diverse researchers.

Finally, a performance evaluation of plasticity of user interface was conducted: it is concluded that the use of dynamic techniques can enhance the user experience to a much greater extent than more basic approaches, although optimisation of usability parameter trade-offs needs further attention.

It is noted that most of the work reviewed originated from a limited range of cultural perspectives. To make an interface simultaneously usable for users from a diverse range of cultural backgrounds will require a very large amount of adaptation, but the powerful principles of plasticity of user interface design hold the future promise of an optimum tool to achieve cross-cultural usability.

提出了关于通用性，用户界面设计的可塑性以及促进具有通用性的界面开发的研究综述。这项调查基于跨越五十五年的165篇研究论文。介绍了自适应或智能用户界面（AUI或IUI）的基础，重点关注三个核心领域：人工智能（AI），用户建模（UM）和人机交互（HCI）。为了比较各种AUI，给出了建议的分类法。一个结论是，必须有一个有效的训练向量来实现机器学习算法的快速最佳收敛，但这的关键是数据集的边界，目标是要获得准确的用户偏好模型，必须从从人类互动中获得的有限数量的数据集。

随着用户的全球移动性，界面设计必须考虑到用户的能力和文化，这些能力和文化源于实际的用户行为，而不是基于他们的反馈。关键问题是接口是应在系统控制下自适应还是应在用户控制下自适应。确定了对以下需求的“功能组件”：存储有关最终用户的先验信息；“推论组件”确定用户界面实际需要适应的程度；以及“效率组件”实际确定对用户界面的适应程度。适应性无缝地应用于系统。应用电子学习是当务之急：使用机器智能来实现适当的可学习性是理想的，最好通过“娱乐性互动”来增强。适应的普遍应用在于未来，但在忽略将使用AUI的系统的其他参数时，无法确定AUI属性。必须更全面地了解人类的心理模型，这需要高度多学科的方法以及不同研究人员之间的合作。

最后，对用户界面的可塑性进行了性能评估：结论是，尽管需要进一步优化可用性参数，但是使用动态技术可以比更基本的方法更大程度地增强用户体验。

应当指出，所审查的大多数工作都源于有限的文化视角。为了使界面能够同时适用于来自不同文化背景的用户，需要进行大量的调整，但是用户界面设计可塑性的强大原理为实现跨文化可用性提供了一种理想的工具。