Human emotions, through sensory engagement and the perception of a technology, strongly influences the use of that technology (Berni & Borgianni, 2021). This focus on the human experience is at the core to User Experience (UX) and manifest when developing the tangible aspects of design such as utility, usability, function, and interface (Berni & Borgianni, 2021). To highlight some specific examples, UX is evident in physical products through its tactile feel, size, or weight, in software through its navigation, speed, or accessibility, in services through its flexibility or perks, and in all products through its feedback mechanisms, marketing, or intuitiveness.
Image generated by Gemini Flash 2.5 (Google, 2025), with the prompt sequence outlined below.
As of 2025, generative artificial intelligence (AI) can be mistaken as human (Jones & Bergen, 2025); that is, a human-AI interaction may be experienced as a human-human one. For millennia, some human-human interactions have been fraught with mistrust, misunderstandings, mismanagement, and miscommunications, and these very same issues have been recognized in the UX of AI systems (Wang et al, 2019). Therefore, the challenge for high-consequence AI systems is to navigate the often irrational way humans reason – heuristic biases and cognitive shortcomings – in order to adequately explain its thoughts and actions (Wang et al, 2019).
Wang et al. (2019) propose a framework to address the nuance of human reasoning that AI systems should employ. These include the mitigation of representative bias, availability bias, anchoring bias, confirmation bias, and the moderation of trust (Wang et al, 2019).
While AI tools are still in a relatively early stage of development, they are being adopted and found as useful. Student in higher Ed are drawn to AI use for the expectation of utility, enjoyment, and habitual use (Strzelecki, 2024). A multitude of studies found AI feedback of student writing to be beneficial to student performance and as a valid and reliable yet still with room for improvement compared to human evaluators (Kuzminykh et al., 2024).
As I reflect on my generative AI use, I would use the term ‘playful’ – I find it explorative, low consequence, and I enjoy using it. Similar to what Wang et al (2019) identify, I hesitate to blindly use AI in higher consequence circumstances.
Prompt Sequence
Generative AI model used: Gemini 2.5 Flash
Method: Prompt Chaining:
- Image text prompt creation and iteration
- Text – image creation and iteration
Initial prompt:
Can you act as a prompt engineer who is an expert of editing text to be utilized in a text-to-image prompt. I will give you the first draft of a textual prompt, and you will output refined text to best visualize the description. Clarify first.
Please create an image of an emotionally tense stand-off between two individuals. They are longing to connect, yet something is holding them back. One of the person is in the centre of the frame and their head is just a black box, the other person is a human viewed in third person. Make it in the style of an impressionism oil painting with a painteresque vignette. Clarify first
Final Image prompt:
Impressionism oil painting, emotionally tense stand-off between two individuals. They long to connect, but something holds them back. The central figure is a man with a black box for a head, who is positioned further back in the frame. The other figure is a woman, viewed in third-person in the foreground. The woman is trepidatious and apprehensive, yet with some interest, and an overall air of mistrust, engaging with the man but remaining reserved. The painting is set on a wet, rainy city street, reflecting the moody and atmospheric light of an overcast day.
Biases in the imagery and AI prompt development:
The more ‘pushy’ individual with a black box head became a man wearing a suit (i.e. emphasis on a man and a suit symbolic of business, potentially extrapolating to a man=business). There may be a lot to unpack in the gender choice of the AI to being a man and a woman and the tension that exists between them from a traditional strength=power role and from courtship to just label a few. The imagery appears to be reminiscent of Europe, though that may be an artifact of the impressionism style which did originate in France. Both characters are white – while race has been identified as a continuous issue in AI image generation, this outcome may also be determined by my Eurocentric art style choice of Impressionism.
References
Berni, A., & Borgianni, Y. (2021). From the Definition of User Experience to a Framework to Classify Its Applications in Design. Proceedings of the Design Society, 1, 1627–1636. https://doi.org/10.1017/pds.2021.424
Google. (2023). Gemini [Large language model]. https://gemini.google.com/
Jones, C. R., & Bergen, B. K. (2025). Large Language Models Pass the Turing Test. arXiv. https://arxiv.org/abs/2503.23674
Kuzminykh, I., Nawaz, T., Shenzhang, S., Ghita, B., Raphael, J., & Xiao, H. (2024). Personalised Feedback Framework for Online Education Programmes Using Generative AI (No. arXiv:2410.11904). arXiv. https://doi.org/10.48550/arXiv.2410.11904
Strzelecki, A. (2024). To use or not to use ChatGPT in higher education? A study of students’ acceptance and use of technology. Interactive Learning Environments, 32(9), 5142–5155. https://doi.org/10.1080/10494820.2023.2209881
Wang, D., Yang, Q., Abdul, A., & Lim, B. Y. (2019). Designing Theory-Driven User-Centric Explainable AI. Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems, 1–15. https://doi.org/10.1145/3290605.3300831