The text below is a brief excerpt from my final thesis in sociology under the title "‘AI’ in situated action. Unlocking the role of UX and UI design in enabling ‘AI’ in everyday life". (Based on empirical data from a research study I conducted.)

On matters of trust

Trustworthy AI is not only a category defined in the EU AI Act (EU 2024; HLEG AI 2019). Trust and transparency are also central concepts in the field of UX and UI design. This is not explicitly considered in regard to AI. However, as AI is increasingly incorporated into the UX and UI of products and services, these concepts have become relevant to the process of integration. They have been and continue to be institutionalized as principles in the field and as standards in business. 
[...] 
What does this imply in practice?
[...] 
When it comes to matters of trust or matters of intimacy in regard to AI, there are various possible approaches for analysis. As my results have highlighted, one practice in the field of UX and UI design to enable intimacy in interactions with AI (also as a means for trust) is to anthropomorphize these products through persona or voice design. Another possibility, to enable trust, is to generate value. This could be done by offering benefits. For example, through the transparent presentation of sources (characterized by reliability), or through personalized support and confirmation (characterized by sociality). 
[...] 

Meaning is negotiated in interactions

As Blumer (1962:190) stated: “People – that is, acting units – [...] act toward situations.” They also act toward things based on the meaning they have for them. This meaning is negotiated in interactions (Blumer 1969) that are both situational and situated (Goffman 1963). Furthermore, meaning can be used and changed in interpretative processes (Blumer 1969). Consequently, meaning attributed to entities such as AI is inherently relational and procedural. 
[...]

Trustworthiness is not an inherent property of AI itself

Design elements and decisions, such as anthropomorphization, personalization, the representation of corporate status through, for example, the tone of voice of a chatbot, or added transparency, not only influence the outcome of how AI is perceived and evaluated (whether or not it is a trustworthy actor), but also change the manner of the actual interaction. The idea of AI as trustworthy is made possible by integrating AI into a broader context, which means “by acts of contextualization and recontextualization”(Amrute 2025:4). It is not an inherent property of AI itself. However, meanings such as trustworthiness are negotiated in the interaction itself. Co-dependent and co-produced. And this process is situational and situated. Overarching contexts (meso-level) and individual experiences (micro-level) are interconnected. 
[...]

Information presented as transparent is always embedded in an interaction that has been pre-framed 

Trust and intimacy are seamlessly intertwined. This implies, for example, that information presented as transparent (in an attempt to counteract the opacity of AI) is always embedded in an interaction that has been pre-framed. This kind of interaction “creates a space where information becomes an intimate, personalized medium” (Bozdağ 2025:2300). This is all the more relevant to consider if these interactions are more like an “echo chamber of affection” (Mahari/Pataranataporn 2024). 
[...]

Accepting something or someone as trustworthy is not equivalent to being trustworthy

Existing cultural concepts of trust and intimacy should also be taken into account. What is considered trustworthy in society (on a macro level)? Accepting something or someone as trustworthy is not equivalent to being trustworthy. For example, if people tend to trust AI with a pleasant voice more, that does not mean the AI is actually trustworthy (Reinhardt 2023:740). Furthermore, the question arises as to whether the concept of trustworthy AI or the attribution of trustworthiness to a technology is based on a fundamental conceptual problem (see Coeckelbergh/Metzinger 2020). Some consider it “ethical washing” (Freiman 2023:1352) or a “bedtime story for tomorrow's customers […] [and] first and foremost, [a] conceptual nonsense” (Metzinger 2019). Therefore, it is key to consider and inquire about what is actually being made possible, how, and why.

If you’d like to find out more (about my findings) or discuss any specific points, please don’t hesitate to get in touch!