<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Shenyue Jian on Medium]]></title>
        <description><![CDATA[Stories by Shenyue Jian on Medium]]></description>
        <link>https://medium.com/@jiansy00?source=rss-6b674e101aef------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*ZPnOSYP_aeAqUNWo</url>
            <title>Stories by Shenyue Jian on Medium</title>
            <link>https://medium.com/@jiansy00?source=rss-6b674e101aef------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 05:12:20 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@jiansy00/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Mapping Everyday AI Use Across University Support Work]]></title>
            <link>https://medium.com/@jiansy00/mapping-everyday-ai-use-across-university-support-work-3b5a4b0f166c?source=rss-6b674e101aef------2</link>
            <guid isPermaLink="false">https://medium.com/p/3b5a4b0f166c</guid>
            <dc:creator><![CDATA[Shenyue Jian]]></dc:creator>
            <pubDate>Sat, 30 May 2026 21:24:51 GMT</pubDate>
            <atom:updated>2026-06-04T03:18:42.380Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Blog 4 (This blog post documents my major project journey for the MA Service Design course at LCC, UAL.)</em></p><p>Over the past few weeks, my project has shifted to how AI is actually used by administrative staff within London College of Communication (LCC) at University of the Arts London. While there are extensive researches on AI adoption across industries, there is far less written about how administrative and support teams in higher education engage with these tools day to day. This revealed an interesting gap, the internet discussions often focus on what AI could do, but my project is becoming increasingly concerned with what people are actually doing.</p><p>There are some recurring themes appear in my secondary research.</p><p>Many studies position AI as a automation tool for repetitive administrative work such as scheduling, organising information, drafting routine communications, and summarising large amounts of content. Researches also highlighted AI’s ability to process large amount of information in very short time than it would take for human researcher. At the same time, UAL’s publicly available guidances remains largely focused on teaching and learning, only very limited amount of information aim for operational staff use. Meanwhile, under UK GDPR regulations, the use of external AI systems with student information has significant legal and ethical concerns, which helps explain why university staff often approach AI cautiously.</p><p>Rather than relying solely on secondary research, I had conversations with nine different administrative staffs at LCC working in reception, information center, library, academic support, language development, and the Student Union. Because these conversations took place within a single institution and some participants discussed organisational practices, I have decide to anonymise participants unless they have explicitly agreed otherwise.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mu167fnaJfc_cEkeE0npTQ.png" /><figcaption>Part screenshot of organised insights</figcaption></figure><p>Several strong patterns emerged:</p><p>Firstly, AI is rarely being used to generate content from scratch. Most participants described using AI as a checking tool rather than a creating tool. Grammar checks, cohesion checks, summarising long documents, extracting references, and reviewing writing against requirements were all considered acceptable uses.</p><p>Secondly, trust remains a significant issue. Staff repeatedly described AI as something that requires verification. Outputs are checked, reviewed, and treated cautiously. Several participants specifically mentioned concerns about hallucinations, misinformation, and risk of over reliance. This closely reflects Dogru and Krämer research (2025) on appropriate reliance in human and AI interaction, which argues that effective AI use sits between two extremes: over reliance, where users blindly accept incorrect outputs, and under reliance, where useful outputs are rejected due to distrust.</p><p>Thirdly, many participants viewed AI primarily as a tool, a source. Usages of creative generation, image generation, and writing entire pieces of work are considered unacceptable, particularly within an arts university context. Several participants drew very clear boundaries between using AI to support existing work and allowing AI to replace human judgement or authorship.</p><p>I also began mapping where AI appears across different stages of work, a pattern of selective adoption emerged. Staff described different levels of comfort depending on where AI entered the workflow. AI was most accepted Before and After a task. Resistance happen more in the During stage which relied heavily on human judgement, interpretation, or personalised communication. Rather than replacing work, AI currently appears to function as a supporting layer around work, helping staff prepare, review, and refine outputs while leaving core decision making largely in human hands.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*KRSkEFIv_Z-qUXs6LIuFbg.png" /><figcaption>AI Usage in Work Stages</figcaption></figure><p>Alongside this, I have started mapping the organisational structure surrounding support services. Interestingly, many staff members were unsure how different departments connected across UAL or where their teams sat within the wider organisational structure. This uncertainty suggests that understanding organisational relationships may be just as important as understanding AI tools.</p><p>At this stage, several questions remain:</p><ol><li>Should this project focus specifically on LCC or expand to UAL more broadly?</li><li>Should I focus primarily on Microsoft Copilot, given its institutional approval?</li><li>Should my intervention sit within individual staff practice, team-level support, or wider organisational guidance?</li></ol><p>These questions will shape and answered in the next stage of the project.</p><p>What has become clear so far is that the challenge is not whether staff use AI or not. The more interesting question is: how might university support staff build clarity and confidence in appropriate AI use within their daily workflows while maintaining quality, data security, and human judgement?</p><h3>Reference</h3><p>Dogru, E.Ö. and Krämer, N.C. (2025). Investigating appropriate reliance on AI-Based decision support systems: the role of expertise, trust, and self-confidence. <em>Journal of Decision Systems</em>, [online] 34(1). doi:https://doi.org/10.1080/12460125.2025.2593251.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3b5a4b0f166c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[When the Problem Exists but Access Doesn’t]]></title>
            <link>https://medium.com/@jiansy00/when-the-problem-exists-but-access-doesnt-cb76f31b1aee?source=rss-6b674e101aef------2</link>
            <guid isPermaLink="false">https://medium.com/p/cb76f31b1aee</guid>
            <dc:creator><![CDATA[Shenyue Jian]]></dc:creator>
            <pubDate>Fri, 29 May 2026 21:00:47 GMT</pubDate>
            <atom:updated>2026-06-04T03:18:30.730Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Blog 3 (This blog post documents my major project journey for the MA Service Design course at LCC, UAL.)</em></p><p>As my project direction became more focused around the insurance industry, the term “frontline insurance work” needs more clarification. Many insurance workers operate behind screen with large volumes of information processing, they handle direction decisions, communication, and accountability within the service.</p><p>Below are some frontline insurance roles I identified:</p><h4><strong>Underwriters</strong></h4><p>who decide whether someone should be insured and at what cost</p><h4><strong>Claims handlers</strong></h4><p>who manage customer cases after something has gone wrong</p><h4><strong>Customer service and case handlers</strong></h4><p>who act as the bridge between customers and the company</p><h4><strong>Compliance checkers</strong></h4><p>who ensure decisions follow legal and ethical regulations</p><h4><strong>Risk assessment and fraud review staff</strong></h4><p>who evaluate suspicious or high risk claims</p><h4><strong>Claims operations staff</strong></h4><p>who monitor workflows, delays, and processing systems</p><p>The roles are different, but all of them involve large amounts of repetitive administration, judgement, decision making, and communication, which makes them highly relevant to current focuses on AI adoption.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wk1e1kZl3Yas9YgdjEHMmg.png" /><figcaption>AI adoption in insurances and workplaces desk research</figcaption></figure><p>As part of desk research, I also began looking at online communities such as Reddit for more personal opinions. Compared to formal reports, these conversations revealed more emotional and practical reactions towards AI in insurance work.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Tx46AksER97JVdWQXuwxdg.png" /><figcaption>Screenshot of Reddit post “What to look for in insurance AI solutions?”</figcaption></figure><p>One recurring pattern was insurers generally found AI useful for reducing repetitive tasks with replacing human judgment. Discussions around quoting tools, document handling, summaries, and call routing frequently described AI as a “significant time saver” (Reddit.com, 2026), especially when integrated into existing workflows.</p><p>At the same time, trust remained a major concern.</p><p>Many workers repeatedly emphasised that insurance decisions cannot rely on “maybe right” outputs. Others stressed that AI still requires human oversight, particularly in high stakes situations involving risk and legal responsibility. This coherent with Human-in-the-loop (Gómez-Carmona et al., 2024) thinking, that AI functions more effectively as support rather than replacing human judgement.</p><p>Another consistent insight was customers still prefer dealing with people, especially for important or stressful situations such as insurance claims. Several users described frustration with automated systems and concerns of AI heavy customer service could make people “hate insurance more”. Reflecting an important limitation in AI adoption, improve in efficiency alone does not guarantee improve service experience.</p><p>Many discussions also suggested that workers were not necessarily resistant to AI itself, but more cautious about where AI should and should not be used. The tools that received the most positive reactions were those supporting repetitive administrative work, such as document processing, summarisation, data entry, and email organisation. In contrast, fully automated decision making systems create much stronger distrust.</p><p>These findings revealed that the issue may not be whether AI exists in the workflow, but whether workers can have the clear boundaries between human judgement and automation. Then, I began reaching out to insurance professionals on LinkedIn to conduct primary research interviews. Initially, I assumed insurance workers would be relatively open to talking about their experiences, similar to my experience back home. However, the reality was more complicated.</p><p>Although 20 people accepted my connection requests, only 3 people responded to messages, and one agreed to a conversation out of 99 invitations sent. Reflecting on this process, I have been constantly refining the wording and framing of my requests, finding people who interested in the topic, and encouraging responses became a research challenge in itself.</p><p>For example, instead of asking directly for collaboration or formal interviews, I gradually shifted towards softer and more open language, such as “short conversations” or “informal discussions”. I also drew on psychology principles when writing invitations to reduce perceived effort, increase autonomy, and highlight mutual benefit.</p><p>The process made me realise that participation itself is part of service design. If people already feel overloaded, uncertain, or cautious about AI in their workplace, asking them to formally discuss it becomes another barrier.</p><p>Eventually, I conducted a phone interview with one insurance professional.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3ASy_6WcbFvjhfFbl-Smzg.png" /><figcaption>Interview questions prepared</figcaption></figure><p>The insights were detailed and highly valuable, however, the participant requested complete anonymity and was not comfortable with ongoing collaboration. While completely understandable, this highlighted a practical challenge for the project. My research suggests the problem genuinely exists, but access to participants and long term collaboration within the insurance sector is difficult, especially within project’s given timeframe.</p><p>Reflecting on this stage of the project, I decided to keep the core themes but reconsider the context in which I investigate in. Issues around AI trust, uncertainty, training gaps, and hidden verification work were also appearing in conversations within the university context. At the same time, the university environment offers much greater accessibility for interviews and potential co-creation opportunities.</p><p>Because of this, I decide to shift my focus towards AI use within university support services and administrative workflows, with this research question:</p><blockquote><strong>How might university administrative staff engage with AI tools ethically for everyday workflow support?</strong></blockquote><h3>References</h3><p>Gómez-Carmona, O., Casado-Mansilla, D., López-de-Ipiña, D. and García-Zubia, J. (2024). Human-in-the-loop machine learning: Reconceptualizing the role of the user in interactive approaches. <em>Internet of things</em>, [online] 25, pp.101048–101048. doi:<a href="https://doi.org/10.1016/j.iot.2023.101048">https://doi.org/10.1016/j.iot.2023.101048</a>.</p><p>Reddit.com. (2026). <em>Reddit — Please wait for verification</em>. [online] Available at: <a href="https://www.reddit.com/r/InsuranceAgent/comments/1o639br/comment/njdrz5e/">https://www.reddit.com/r/InsuranceAgent/comments/1o639br/comment/njdrz5e/</a> [Accessed 29 May 2026].</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cb76f31b1aee" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AI Adoption and the Readiness Gap: Shifting Focus Beyond Design]]></title>
            <link>https://medium.com/@jiansy00/ai-adoption-and-the-readiness-gap-shifting-focus-beyond-design-89b334ecc1ed?source=rss-6b674e101aef------2</link>
            <guid isPermaLink="false">https://medium.com/p/89b334ecc1ed</guid>
            <dc:creator><![CDATA[Shenyue Jian]]></dc:creator>
            <pubDate>Mon, 25 May 2026 14:07:37 GMT</pubDate>
            <atom:updated>2026-06-04T03:18:18.373Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Blog 2 (This blog post documents my major project journey for the MA Service Design course at LCC, UAL.)</em></p><p>As my research developed, it became clear that the challenges I observed in design were part of wider pattern. Across industries, AI is being introduced into workflows at a speed that organisations and workers are not fully prepared for. Research highlights a significant gap between adoption and readiness:</p><blockquote>97% of UK organisations report at least one AI skills gap (Gardiner &amp; Theobald, 2025).</blockquote><blockquote>83% of leaders use AI regularly, but only 21% of frontline workers receive adequate training (Brown, 2024).</blockquote><p>This suggests that the issue is not access to AI itself, but the lack of support, structure, and clarity around how it should be used in practice. While there are controlled experiments showing use of AI improves in speed and quality, real-world experiences are more mixed. For example, AI can increase productivity by up to 33% in certain conditions (Nielsen, 2023), yet 40% of workers report that it saves them no time at all (Skibba, 2026). Additionally, the issues of workslop is great representation of AI does not automatically increase productivity.</p><p>The expectation and reality gap became a key focus in my research, based on research, three core issues emerging across sources are:</p><h4><strong>Training Gap</strong></h4><p>Workers are expected to use AI without sufficient time, guidance, or structured learning support.</p><h4><strong>Trust Gap</strong></h4><p>AI output is not fully trusted, requiring constant verification.</p><h4><strong>Autonomy Gap</strong></h4><p>Instead of empowering workers, AI could increase pressure and reduce control over how work is done.</p><p>To better understand where these issues become most critical, I expanded my research into multiple industries using methods learned from the Design Futures unit, including evidence safari and horizon scanning. Through this process, the insurance industry emerged as a particularly strong focus area.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8dK333WjkgAscWUegPTR-Q.png" /><figcaption>Insurance direction project mapping</figcaption></figure><p>Roots Research (2025) shows that:</p><blockquote>Over 80% insurance executives identified AI as a top strategic initiative for 2025.</blockquote><blockquote>70% of insurance companies are actively testing or exploring AI solutions, fewer than, however only 22% have advanced their AI projects from the testing phase to full production.</blockquote><p>These suggests that adoption alone is not enough.</p><p>Insurance is a highly regulated environment, where decisions have financial and legal consequences. Unlike design, workers cannot rely on experimentation or informal learning to the same extent. AI is often being layered onto existing systems without redesigning the workflow itself.</p><p>As noted in my earlier research mapping:</p><blockquote><em>Workers like claims handlers and underwriters are now doing their old job plus checking AI output with their own judgment, and additional accountability to take. The work hasn’t got easier, it’s got murkier.</em></blockquote><p>This creates uncertainty and increases cognitive load, rather than reducing it.</p><p>To investigate this further, I plan to use the following methods to research among frontline insurance workers:</p><ul><li>Semi-structured interviews with claims handlers and underwriters</li><li>Workflow mapping to identify where AI is used or avoided</li><li>Stakeholder interviews to compare frontline and managerial perspectives</li></ul><p>Supported by key theoretical frameworks:</p><h4><strong>Technology adoption theory</strong></h4><p>Suggests new technology adoption depends on users’ perceived usefulness and ease of use.</p><h4><strong>Human-in-the-loop thinking</strong></h4><p>Frames AI as a collaborative system requiring human judgement at key decision points.</p><h4><strong>Cognitive load theory</strong></h4><p>Explains how managing AI outputs alongside existing responsibilities can lead to overload and fatigue.</p><p>Which all land me to current project direction:</p><blockquote><strong>How might we help frontline insurance workers develop confident judgment about when and how to use AI, so reduce their burden and match top-down expectations?</strong></blockquote><p>The next steps will focus on conducting primary research with insurance workers and stakeholders, testing assumptions around workflow and AI use, and identifying where meaningful intervention could reduce uncertainty, cognitive strain, and verification burden.</p><p>AI has the potential to improve productivity and decision making, but without structure, training, and clarity, it can just as easily create confusion, extra work, and reduced trust. The issue is not whether AI works, but whether people are supported to use it properly.</p><h3>References</h3><p>Brown, L. (2024). <em>AI skills gap emerging between managers and frontline workers, study suggests</em>. [online] <a href="http://www.peoplemanagement.co.uk./">www.peoplemanagement.co.uk.</a> Available at: <a href="https://www.peoplemanagement.co.uk/article/1878798/ai-skills-gap-emerging-managers-frontline-workers-study-suggests">https://www.peoplemanagement.co.uk/article/1878798/ai-skills-gap-emerging-managers-frontline-workers-study-suggests</a> [Accessed 7 May 2026].</p><p>Gardiner &amp; Theobald (2025). <em>AI Labour Market Survey</em>. [online] p.3. Available at: <a href="https://assets.publishing.service.gov.uk/media/6960ef384343a0da370869b7/AI_Labour_Market_Survey_2025_report.pdf">https://assets.publishing.service.gov.uk/media/6960ef384343a0da370869b7/AI_Labour_Market_Survey_2025_report.pdf</a> [Accessed 7 May 2026].</p><p>Nielsen, J. (2023). <em>AI Helps Elite Consultants: Higher Productivity &amp; Work Quality, Narrower Skills Gap</em>. [online] Jakob Nielsen on UX. Available at: <a href="https://jakobnielsenphd.substack.com/p/ai-elite-consultants-productivity">https://jakobnielsenphd.substack.com/p/ai-elite-consultants-productivity</a> [Accessed 7 May 2026].</p><p>Roots (2025). <em>State of AI Adoption in Insurance 2025</em>. [online] Available at: <a href="https://www.roots.ai/hubfs/Reports%20and%20Whitepapers/Roots%20-%20State%20of%20AI%20Adoption%20in%20Insurance%202025.pdf">https://www.roots.ai/hubfs/Reports%20and%20Whitepapers/Roots%20-%20State%20of%20AI%20Adoption%20in%20Insurance%202025.pdf</a> [Accessed 25 May 2026].</p><p>Skibba, R. (2026). <em>Bosses say AI boosts productivity — workers say they’re drowning in ‘workslop’</em>. [online] the Guardian. Available at: <a href="https://www.theguardian.com/technology/2026/apr/14/ai-productivity-workplace-errors">https://www.theguardian.com/technology/2026/apr/14/ai-productivity-workplace-errors</a> [Accessed 6 May 2026].</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=89b334ecc1ed" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Human Dividend: Reimagining Work Through a Feminist Lens]]></title>
            <link>https://medium.com/@jiansy00/human-dividend-reimagining-work-through-a-feminist-lens-60cefe7f9fff?source=rss-6b674e101aef------2</link>
            <guid isPermaLink="false">https://medium.com/p/60cefe7f9fff</guid>
            <dc:creator><![CDATA[Shenyue Jian]]></dc:creator>
            <pubDate>Wed, 13 May 2026 21:37:01 GMT</pubDate>
            <atom:updated>2026-05-13T21:45:33.016Z</atom:updated>
            <content:encoded><![CDATA[<p>As one of many women, the workplace is where I will step into after graduating from university. In fact, I had already experienced it before joining this course. From what I observed, women workers are more likely to experience various ‘life incidents’ compared to men, such as period pain, maternity leave, and childcare responsibilities. Women’s bodies go through cycles that can negatively affect our position within competitive, productivity driven workplaces.</p><p>Back in my home country, female job seekers are often asked directly about their relationship status or whether they plan to have children. Women already in work would carefully plan around these cycles to minimise potential impacts on promotion opportunities, pay rises, or job security. If a woman goes through pregnancy, worries quickly follow:</p><ul><li>Who is going to take my role?</li><li>Will my position still exist when I return?</li><li>Will I be able to balance childcare with work?</li></ul><p>A UK-based survey found that 29% of women experienced unfavourable treatment at work related to pregnancy or pregnancy-related health issues (Psarros, 2026, p.10), with different forms of treatment shown in Figure 1.</p><figure><img alt="Psarros, A. (2026b). Types of unfavourable treatment. [Table] Maternity Action, p.11. Available at: https://maternityaction.org.uk/wp-content/uploads/UNISON-maternity-rights-survey.pdf [Accessed 13 May 2026]." src="https://cdn-images-1.medium.com/max/1024/1*O39jrnp_MHHsPBcNYmpNGg.png" /><figcaption>Figure 1: Psarros, A. (2026). <em>Types of unfavourable treatment</em>. [Table] <em>Maternity Action</em>, p.11. Available at: <a href="https://maternityaction.org.uk/wp-content/uploads/UNISON-maternity-rights-survey.pdf">https://maternityaction.org.uk/wp-content/uploads/UNISON-maternity-rights-survey.pdf</a> [Accessed 13 May 2026].</figcaption></figure><p>At the same time, women are still expected to carry out housework. Analysis from the Office for National Statistics shows that women undertake, on average, 60% more unpaid work than men (Office for National Statistics, 2016). Overall, women often have less time and energy to focus solely on work.</p><p>As an MA Service Design student, I had the opportunity to work on the Design Futures unit, focusing on women at work through a feminist lens. This blog documents the design journey and explores two questions:</p><ul><li>What does it mean to design futures through a feminist lens?</li><li>What difference does that lens make?</li></ul><h3>Brief Focus</h3><p>Traditionally, much of the work women do has remained invisible and undervalued. Gender discrimination within workplaces often happens in silence, appearing in such familiar ways that many people no longer question it until they personally experience it. This brief invited us to imagine a future model of work centred around care, supporting women through different stages of life while recognising and validating all forms of labour.</p><p>In partnership with PILLA, the project explored possible future shifts in work experiences, workplace needs, and wider changes in policy, culture, and design that could contribute towards a better working life.</p><p>Feminist speculative design and critical design theories were also introduced throughout the unit. Prado de O. Martins argues that anyone attempting to criticise society can easily sound patronising, when speaking from a position of privilege, it becomes easy to overlook issues surrounding race, class, and gender (2014, p.4). This became an important consideration of not ignoring the detailed causes of why and how it happens in what context when critiquing systems through a feminist lens. This prompted an important question throughout the project:</p><blockquote>How should we practise feminist perspectives within design?</blockquote><h3>What the Feminist Lens Made Visible</h3><p>Critical service design challenges existing system assumptions and creates space to imagine alternative futures. Drawing on Salinas, Yarrow, and Lagedamont (2024), it aims to better understand current systems before materialising fictional future services that provoke discussion around preferable futures.</p><p>Looking through a feminist lens requires us to make entirely different assumptions. Rather than focusing purely on efficiency or productivity, we began questioning whose bodies, whose time, and whose lifestyles had originally been designed into the workplace.</p><p>Personally, it was quite challenging to shift into the mindset of reimagining the future, as my previous experiences relied more on logical and analytical thinking. One early activity became an effective icebreaker. We were asked to imagine what a feminist city might look like in the future. Our group started from a familiar context: London in 2036. Firstly, we identified everyday women’ s inconveniences, where subtle discomfort happens, then amplifying them into futuristic service ideas (Figure 2).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fjECAGle2w6EAQJTe4Q_7A.jpeg" /><figcaption>Figure 2: London in 2036 draft</figcaption></figure><p>This activity immediately drew our attention to areas often ignored or covered by the dominant systems.It encouraged us to consider the additional burdens women experience across different cultures, locations, and stages of life. As a woman myself, I strongly resonated with most issues we identified. When ideating through a feminist lens, ideas can sometimes become intentionally utopian or extreme. This creates space to move beyond current societal expectations and opens up possibilities for imagining something new. It was an inspiring process, sparking meaningful conversations within our group and across the class, and offering valuable ways of reframing how we think.</p><p>Evidence safari and horizon scanning research methods were introduced, both formed part of desk research approaches, but with slightly different focuses. Evidence safari concentrated on identifying existing behaviours, systems, and patterns, while horizon scanning built upon broader research to identify emerging shifts, signals, or events. The combination of both help create a more holistic understanding of current systems and identify leverage points for future intervention. First mapping workshop in class familiarised us with these research methods (Figure 3).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jYlvYXS3h8P7oJNaEXXYtg.png" /><figcaption>Figure 3: Evident Safari and Horizon Scanning workshop in class</figcaption></figure><p>Sixteen thematic boxes mapped information our group gathered that related to women and work (Figure 4). At surface level, many trends suggested workplaces were gradually becoming more fair and accommodating. Employers began to provide flexible working policies for carers, hybrid working was more widely acceptable, and organisations were beginning to adopt “ethical AI frameworks” to reduce gender bias in recruitment and promotion systems.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QlJszPa0Q4sO8aZurFUQTg.jpeg" /><figcaption>Figure 4: Group sixteen thematic boxes mapping</figcaption></figure><p>These signals suggested a future where work appears more inclusive. However, the feminist lens pushed us to ask another question:</p><blockquote>Accommodating for who?</blockquote><p>The care economy brought to our attention. The global caring market has valued at trillions of dollars, yet much of this labor is unpaid, invisible, and disproportionately undertaken by women. This contradiction sat within the economic box of our mapping. On the surface, care work is framed as personal responsibility rather than economically valuable labour. But through a feminist lens, the seriousness of this contradiction became impossible to ignore.</p><p>Technology is changing and adapting, AI and automation are not in a completely neutral position, it more likely to replace jobs involving structured and repetitive tasks (Azpúrua, 2026). At the same time, the skills least likely to be replaced are human capabilities such as emotional intelligence, relational care, and navigating complex situations (Office for National Statistics, 2025, p.13). Ironically, these are often the same skills developed through caregiving responsibilities, skills historically undervalued within professional environments.</p><p>This contradiction became our main focus. If AI eventually automates the systematic and repetitive aspects of work, then what remains is fundamentally human. Therefore, the question is no longer simply how women return to work after career breaks, but whether the invisible skills developed through care can finally become recognised as legitimate professional value.</p><p>Ahmed (2017) argues that feminist theory begins not from abstract or academic distance, but from experiences of the world failing to accommodate us. This perspective strongly resonated throughout the project. If we had only conducted horizon scanning without feminist framing, the future might have appeared optimistic and progressive. The feminist lens did not change the trends themselves, but it completely changed how we interpreted what those trends were actually evidencing.</p><h3>Researching Through Making</h3><p>Based on our research, we began asking:</p><blockquote><strong>What if the future of work was no longer centred around productivity, but around care?</strong></blockquote><p>We started constructing a speculative world by imagining the kinds of futures we would genuinely want to live in (Figure 5). Our imagined future became a feminist social democracy world where care functions as the primary form of social value, not profit. Power is distributed instead of concentrated. Communities have stronger voices in shaping policies and systems. Women and carers hold equal political and economic influence, and policy is designed with them, rather than for them. Care means more than healthcare or childcare. It means people are able to choose how they live freely and feel supported not only by institutions, but by each other.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8bzUeodC3Z80LyWIh8bJ5A.png" /><figcaption>Figure 5: World building ideation</figcaption></figure><p>One of the most valuable lessons I learned during this unit was the concept of research through design. Discussing speculative futures purely through imagination or abstract concepts can often feel hard to understand for others. Having physical prototypes to give out and interact with, even rough artefacts make conversation starts more easily, ideas more understandable and critiquable.</p><p>After defining our future world, we brainstormed a range of rapid speculative prototypes (Figure 6) that could provoke discussion, and be tested with participants who had experience applying for jobs or working professionally, regardless of gender. Since our imagined future aimed to benefit everyone, restricting testing only to women felt limiting. Due to time constraints, we focused on two prototypes related to recruitment processes.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*UsRmTGmAT9ulGglw1UMU1A.png" /><figcaption>Figure 6: Prototype idea brainstorming</figcaption></figure><h3>The Life CV</h3><p>The first prototype was Life CV (Figure 7). Compared with traditional CVs in 2026, where career gaps are often hidden or disguised, the Life CV imagined a future where care experiences are translated into recognised professional capabilities. Time spent raising children, caring for family members, recovering from illness, or navigating personal challenges will no longer be gaps, but developing valuable human skills such as sustained attention, emotional resilience etc.</p><p>We printed two CVs and asked participants open questions such as:</p><ul><li>How do you feel about these two CVs?</li><li>What do you think about the way we turn care into experiences?</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*j120hoBTJljU8W7AFZ8BDQ.png" /><figcaption>Figure 7: 2026 resume VS 2036 resume</figcaption></figure><p>Initially, feedback was overwhelmingly positive. We then quickly realised our questioning was unintentionally leading participants towards approval. To create more critical discussion, we reframed the question by assigning participants roles:</p><ul><li>Please consider these CVs from an HR perspective. What values does the 2036 CV offer, and what concerns you might have?</li></ul><p>The responses changed immediately.</p><p>One participant preferred the 2025 CV with the unexplained career gap because the 2036 version felt harder to scan quickly. Reflecting on our design, the care sections are visually different from the rest of the document, unintentionally flagged difficulty rather than strength.</p><p>Another male participant also preferred the 2025 version with an empty career gap. His response reflected something we had already identified in earlier research: workplace cultures are so used to undervaluing care that making care visible itself feels uncomfortable.</p><p>Participants also questioned the rigid formatting of the “care skills” sections (Figure 8). One pointed out that it focused “too much on direct job relevance”, suggesting that care still had to prove itself through corporate language before being taken seriously. Another participant suggested these experiences should be framed more as transferable human strengths rather than checklist style competencies.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ncYa_xGGbmlIteU582iAEw.png" /><figcaption>Figure 8: Life CV Interviews</figcaption></figure><p>A deeper issue also emerged:</p><blockquote><strong>Who validates care experience?</strong></blockquote><p>Unlike professional employment, there won’t be manager recommendations or performance reviews from children. There aren’t credit systems that can validate professional care experience, but our CV format assumes that.</p><p>This tension strongly connected to Auger’s argument that successful speculative design must balance familiarity with provocation, so that it is novel enough to inspire and familiar enough to engage people (2013). Our first version leaned too heavily towards familiarity, care experiences inserted into traditional CV structures, looks clean and readable, but failed to truly challenge the assumptions underlying recruitment itself.</p><h3>Immersive VR Interviews</h3><p>The second prototype explored immersive interviews. In 2036, job interviews could take place as live simulations within dedicated VR centres where candidates are assessed on their soft skills, decision making and immediate responses in AI personalised workplace scenarios. Using Claude, we generated a clickable prototype demonstrating the journey from receiving an interview invitation, simulation, and qualification report with AI suggestion (Figure 9).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-XHljxRauSr6xU3Cwj5D7g.png" /><figcaption>Figure 9: Claude generated interview journey flow</figcaption></figure><p>The first round of testing revealed important insights. Initially, we introduced the prototype without much context and simply asked participants how they felt about it. Participant responses remained quite surface-level and positive, mainly focusing on how AI could help assess people beyond traditional CVs. We then started introducing the broader speculative world first, the conversations became much more critical and reflective. Participants appreciated how immersive assessments could potentially evaluate a wider range of skills beyond CVs. They also felt that the more immersive the experience became, the harder it would be for people to perform or lie about who they are.</p><p>These reflections show that the prototype itself lacked enough justification. Participants struggled to understand without stronger context added. Feedback from tutors emphasised:</p><blockquote><strong>“Show, don’t tell.”</strong></blockquote><p>The artefacts themselves needed to communicate the logic of the future world rather than relying on explanation from us. This feedback fundamentally changed how we approached later testing sessions. Instead of seeking approval, we focused on what question they asked, what felt uncomfortable or unrealistic, what assumption they challenged, not approval.</p><h3>Refining the Future World</h3><p>In the second iteration of the Life CV, care and employment experiences became separated rather than forced into one timeline (Figure 10). Additional features such as care-related awards on the top and community co-signatures on the right corner subtly introduced the values of the future world without explicitly explaining them.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*sR7s6TBFGfbykH8LtupZ4g.png" /><figcaption>Figure 10: Life CV v2 (Left), Co-signed doc (Right)</figcaption></figure><p>The wording also shifted. Instead of framing care as performative corporate language, it became more self reflective and human-centred. These changes replaced the rigid skills blocks, suggesting that care itself was a form of meaningful life experience instead of something requiring justification.</p><p>hooks argues that feminism that only focuses on women accessing existing structure does not change the structure itself (1984, pp. 1–15). This became highly visible through our two CV iterations. The first version inserted care into traditional systems. The second attempted to challenge and challenge the assumptions of the system itself.</p><p>The immersive interview prototype was also updated based on previous findings. In 2036, after CV has been submitted and invited to the VR center, when entering, participant’s baseline emotional data are recorded in neutral conditions. During the VR simulations, AI systems compare candidates only against themselves, not regulated emotional charts. They would be working in realistically simulated environments, AI would identify behaviour patterns, and flag uncertainty.</p><p>Most importantly, human judgement still remains crucial. Everything is recorded for human reviewers and assessed alongside the life CV to consider their suitability. Figure 11 is a <a href="https://ivy-semi-09764800.figma.site/">published prototype</a> of HR supervisory dashboard designed with Figma Make, allowing recruiters to review recordings, annotate moments, and fill in candidate assessment forms for final decision making. Finally, the rejection emails including transparent reasons will be sent out if the candidate is not selected.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*uqOeQws_75Ta6i2pfMu45w.png" /><figcaption>Figure 11: HR supervisory dashboard prototype. Available at: <a href="https://ivy-semi-09764800.figma.site/">https://ivy-semi-09764800.figma.site/</a></figcaption></figure><p>The whole recruitment process flow (Figure 12) prototypes were then tested together as one connected experience. The clear and transparent outcome email was especially appreciated by participants and strongly resonated. One participant commented that:</p><blockquote>“it’s quite nice to see a rejection email and then sort of give reasons why.”</blockquote><p>The inclusion of care experiences within the Life CV was also positively received. One participant responded with:</p><blockquote>“People should work to live, not live to work”</blockquote><p>Reflecting the prototype opened up wider conversations around how success, productivity, and value are measured in current working culture.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*xa3WL0TAZ6Kf6zSPo2EFYg.png" /><figcaption>Figure 12: 2036 recruitment process flow</figcaption></figure><p>Several concerns were raised during testing. Some participants commented that immersive VR interviews could potentially become emotionally triggering or overwhelming under already stressful situations. Another unexpected perspective also emerged, even twelve years later, there are still going to be people who do not believe and reject AI entirely, due to previous experiences of layoffs, replacement, or algorithmic discrimination. This reminded us that AI involved in hiring processes is not automatically neutral. A system intended to reduce one form of exclusion could simultaneously create new forms of exclusion instead.</p><p>Participants also questioned whether people would eventually learn to perform empathy strategically to “game” the system by acting in the most desirable way during assessments. Interestingly, this concern had already surfaced earlier through the “HUMN” drink artefact (Figure 13), designed to artificially enhance empathy, emotional stability, and self-awareness. The feedback reflects how quickly human qualities could become performative once it got measured and rewarded.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/966/1*XBZrCkmwZfv2XYkgnzc_UQ.png" /><figcaption>Figure 13: Gemini generated HUMN drink on Boots Shelf</figcaption></figure><p>These feedback emerged naturally through interactions with the artefacts, not direct prompting from us. It is less about whether the participants “liked” the idea, but more of what assumptions felt uncomfortable, unrealistic, or complicated. Making me feel much closer to what speculative design testing should actually do.</p><h3>Human Dividend</h3><p>Following a decade of AI driven layoffs and algorithmic hiring scandals, the Human Qualities Employment Act 2034 legally required organisations to assess human capability alongside technical skill. Care work, lived experience, and non-linear career paths are formally recognised as professional development. The world is known as the Human Dividend.</p><p>The recruitment process has been redesigned entirely around human-centred values. The Life CV formally included care experiences validated through community verification systems and voice authentication. Recruitment journeys moved through VR assessment centres, human review systems, and transparent decision-making processes.</p><p>Alongside these artefacts, we also created fictional newspaper articles (Figure 14) reacting to the world in different perspectives. Guardian is slightly more optimistic, The Sun considerably less so, reflects that futures are never universally accepted, argument will continue, to the future.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TPrcK9B0v0__nF5D5luAkQ.png" /><figcaption>Figure 14: Newspapers in 2036, “published” by The Guardian and The Sun. Images are ChatGPT and Gemini generated</figcaption></figure><h3>Practice in the Present</h3><p>The future takes some time to come, the process has taught me several lessons to practice in the present.</p><blockquote><strong>Speculative design thinking is challenging yet powerful</strong></blockquote><p>Successful speculative design requires understanding the system we currently exist within before stepping partially out. Futures must feel unfamiliar enough to provoke discussion while remaining grounded enough for people to engage with them meaningfully.</p><p>This balance between familiarity and provocation is something I will continue carrying into future design practice.</p><blockquote><strong>The same signals look different through different lenses</strong></blockquote><p>Feminist theory challenges structures shaped by inequalities of gender, race, and class. It makes invisible systems visible, as hook wrote, centralised the marginalised (1984). This unit taught me how to practice feminist theory into design and into life, it can inform design outcomes, and changes observation itself.</p><blockquote><strong>Feminist thinking also changed team dynamics</strong></blockquote><p>What surprised me most was how feminist frameworks changed how I understood collaboration itself. Design processes often contain invisible hierarchies shaped by who speaks first, whose voice is heard, whose idea gets developed. These subtle signals became much more noticeable through feminist perspectives. Being surrounded by women did not remove power dynamics entirely. Instead, the source shifted towards factors such as confidence, communication style, and design instinct. Sometimes I agreed, argued, or simply followed the flow of discussion. Noticing that was its own feminist lesson.</p><p>This project leaves me with more questions than answers, which perhaps is the point. Speculative design that resolves too cleanly probably was not speculative enough. What I take forward into future practice is the habit of continuously asking:</p><ul><li>Who is this future designed for?</li><li>What is being made invisible?</li></ul><p>Designing the future is not imagining new systems, it is about learning different ways of looking at the present.</p><h3>Reference list</h3><p>Ahmed, S. (2017). <em>Living a feminist life</em>. [online] Durham, Nc: Duke University Press. Available at: <a href="https://assets-us-01.kc-usercontent.com/f7ca9afb-82c2-002a-a423-84e111d5b498/a05011f3-adc3-4fd2-b6b9-d0aa38011636/978-0-8223-6319-4_601.pdf">https://assets-us-01.kc-usercontent.com/f7ca9afb-82c2-002a-a423-84e111d5b498/a05011f3-adc3-4fd2-b6b9-d0aa38011636/978-0-8223-6319-4_601.pdf</a> [Accessed 11 May 2026].</p><p>Auger, J. (2013). Speculative design: crafting the speculation. <em>Digital Creativity</em>, [online] 24(1), pp.11–35. doi:https://doi.org/10.1080/14626268.2013.767276.</p><p>Azpúrua, A.E. (2026). <em>Enhance or Eliminate? How AI Will Likely Change These Jobs</em>. [online] Harvard Business School. Available at: <a href="https://www.library.hbs.edu/working-knowledge/enhance-or-eliminate-how-ai-will-likely-change-these-jobs">https://www.library.hbs.edu/working-knowledge/enhance-or-eliminate-how-ai-will-likely-change-these-jobs</a> [Accessed 11 May 2026].</p><p>hooks, bell (1984). <em>Feminist Theory: From Margin to Center</em>. [online] New York: South End Press , pp.1–15. Available at: <a href="https://funceji.wordpress.com/wp-content/uploads/2017/08/bell_hooks_feminist_theory_from_margin_to_centebookzz-org_.pdf">https://funceji.wordpress.com/wp-content/uploads/2017/08/bell_hooks_feminist_theory_from_margin_to_centebookzz-org_.pdf</a> [Accessed 12 May 2026].</p><p>Office for National Statistics (2016). <em>Women shoulder the responsibility of ‘unpaid work’</em>. [online] Ons.gov.uk. Available at: <a href="https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/earningsandworkinghours/articles/womenshouldertheresponsibilityofunpaidwork/2016-11-10">https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/earningsandworkinghours/articles/womenshouldertheresponsibilityofunpaidwork/2016-11-10</a> [Accessed 9 May 2026].</p><p>Office for National Statistics (ONS) (2025). <em>Gender pay gap in the UK: 2025</em>. [online] <em>Ons.gov.uk</em>, Office for National Statistics, p.13. Available at: <a href="https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/earningsandworkinghours/bulletins/genderpaygapintheuk/2025">https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/earningsandworkinghours/bulletins/genderpaygapintheuk/2025</a> [Accessed 11 May 2026].</p><p>Prado de O. Martins , L. (2014). <em>Privilege and Oppression: Towards a Feminist Speculative Design</em>. [online] pp.1–4. Available at: <a href="https://dl.designresearchsociety.org/cgi/viewcontent.cgi?article=1996&amp;context=drs-conference-papers">https://dl.designresearchsociety.org/cgi/viewcontent.cgi?article=1996&amp;context=drs-conference-papers</a> [Accessed 9 May 2026].</p><p>Psarros, A. (2026). <em>About the rights of pregnant women and new mothers in the workplace</em>. [online] <em>Maternity Action</em>, p.11. Available at: <a href="https://maternityaction.org.uk/wp-content/uploads/UNISON-maternity-rights-survey.pdf">https://maternityaction.org.uk/wp-content/uploads/UNISON-maternity-rights-survey.pdf</a> [Accessed 9 May 2026].</p><p>Salinas, L., Yarrow, L. and Lagedamont, M. (2024). Critical service design for government innovation. <em>DRS2024: Boston</em>, [online] 23–28 June. doi:https://doi.org/10.21606/drs.2024.532.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=60cefe7f9fff" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AI Is Everywhere, But Its Use Remains Unclear: A UX Reflection]]></title>
            <link>https://medium.com/@jiansy00/ai-is-everywhere-but-its-use-remains-unclear-a-ux-reflection-72cdc7834108?source=rss-6b674e101aef------2</link>
            <guid isPermaLink="false">https://medium.com/p/72cdc7834108</guid>
            <dc:creator><![CDATA[Shenyue Jian]]></dc:creator>
            <pubDate>Wed, 06 May 2026 19:27:11 GMT</pubDate>
            <atom:updated>2026-06-04T03:17:46.807Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Blog 1 (This blog post documents my major project journey for the MA Service Design course at LCC, UAL.)</em></p><p>The use of AI has rapidly spread into everyday life. I have observed friends using it to plan travel, compare ticket prices across platforms, and receive food and dietary recommendations based on their location. These tools filling knowledge gaps, enabling faster, more confident decision making.</p><p>Alongside these benefits, it emerging more complex reality. Large organisations have begun reducing headcount based on AI’s anticipated productivity, while remaining employees are increasingly expected to adopt these tools in their daily work. In many cases, this shift occurs with limited guidance or structured training, resulting in work becoming more demanding rather than more efficient (Skibba, 2026).</p><p>A word that captures this issue is <em>workslop</em>, defined as “low-quality output generated by an artificial intelligence system and presented in the workplace as serviceable content”(Collins, n.d.). Rather than reducing effort, workslop often increases review time, extends production processes, and negatively impacts overall performance and morale. This raises questions about whether AI is genuinely improving productivity, and how it can be more effectively integrated into existing workflows.</p><p>Coming from a UX design background, these changes are particularly visible. The rapid emergence of AI tools has introduced a constant pressure to keep up with new technologies, and remain relevant in a job market where expectations keep evolving. Its impact gradually ripple to all of us, especially when the presence of AI skills appears in roles and job descriptions.</p><p>Based on this, my initial focus was to explore how AI is impacting UX designers’ workflows. Research involving 1,478 designers and builders indicates that adoption is already widespread, with 71.1% reporting that they have added AI or gone AI-central in the last 6 months. However, this does not means the integration process is smooth. Trust in AI-generated outputs is still limited, and most designers feel the need to verify results before applying them in practice (UX Tools, 2026). Additionally, the designers’ workflow has shifted, they are no longer solely responsible for creating design, but increasingly prompting AI, curating outputs, fixing errors. With these less visible additional layers of work, certain process becomes faster, but more complex and fragmented overall.</p><p>At the same time, designers do not tend to wait for formal support. Through conversations with designers in my network, I found that many approach AI in a self-directed way, learning through tutorials, experimentation, and build their own understanding through trial and error. This aligns with broader research indicating that AI-related learning is largely informal and embedded within everyday work practices rather than structured training programmes (CBI, 2025).</p><p>This observation led me to question whether UX designers require external intervention at all. Challenges such as tool overload, lack of guidance, and time constraints clearly exist, designers are already actively addressing these issues independently. Also, established UX organisations and training platforms have already started publishing AI-related courses, frameworks, and tools to respond to this demand. For example, offers dedicated courses focused on integrating AI into design workflows, including decision making frameworks that guide when and how AI should be used across different stages of the design process. As a result, there is a risk that any proposed intervention could overlap with existing resources, limiting its relevance or impact.</p><p>Reflecting on this stage of the project, I realised that UX designers represent a group that is relatively adaptable and comfortable with ongoing change. They are used to learning new tools and adjusting their workflows accordingly. Furthermore, this challenge of AI adoption in creative industry are already widely discussed and actively addressed.</p><p>This prompted a reconsideration of my project direction. Rather than focusing on how designers can adapt to AI, I shifted my attention towards identifying contexts where adaptation is more difficult, where the stakes are higher, where people cannot rely on trial and error approaches.</p><h3><strong>Annotated Bibliography</strong></h3><p><strong>Annotation 1:</strong></p><p>CBI (2025). <em>AI Skills Report: Building an AI-Ready Workforce</em>. [online] <a href="http://cbi.org.uk/"><em>cbi.org.uk</em></a>, p.11. Available at: <a href="https://www.cbi.org.uk/media/hekghbfo/cbi-ai-skills-report-final.pdf">https://www.cbi.org.uk/media/hekghbfo/cbi-ai-skills-report-final.pdf</a> [Accessed 6 May 2026].</p><p>This industry report from the Confederation of British Industry examines AI readiness across UK organisations. Highlighted that many AI learning still happen informally, less than half of UK workers report any formal AI-related training. It notes that many businesses experiment with AI without adequate guidance or the capacity to scale what they learn. This source is useful for understanding the institutional barriers to effective AI adoption in the UK workplace context.</p><p><strong>Annotation 2:</strong></p><p>Skibba, R. (2026). <em>Bosses say AI boosts productivity — workers say they’re drowning in ‘workslop’</em>. [online] the Guardian. Available at: <a href="https://www.theguardian.com/technology/2026/apr/14/ai-productivity-workplace-errors">https://www.theguardian.com/technology/2026/apr/14/ai-productivity-workplace-errors</a> [Accessed 6 May 2026].</p><p>This news article explores the growing tension between how AI is perceived by management and how it is experienced by workers in practice. Drawing on survey data from 1,150 US desk workers, it reports that 40% encountered “workslop” within a single month, spending an average of 3.4 hours dealing with it. This adds up to an estimated $8.1 million in lost productivity each year for a 10,000-person organisation. The article also highlights that, despite significant investment in AI and attempts to reduce labour costs, 95% of firms are not yet seeing clear returns.</p><p>This source is useful in grounding the concept of workslop in quantitative evidence, while also showing the gap between expected and actual productivity gains. Although the data is based on a US sample, it still reflects wider patterns in how AI is being introduced into workplaces, and the unintended impact it can have on workers’ experience.</p><p><strong>Annotation 3:</strong></p><p>UX Tools (2026). <em>State of Prototyping Spring 2026</em>. [online] UX Tools. Available at: <a href="https://survey.uxtools.co/spring-2026">https://survey.uxtools.co/spring-2026</a> [Accessed 6 May 2026].</p><p>This open-data survey covers 1,478 designers and builders across 18 regions, looking at weekly tool use, AI adoption, and where designers are investing next. Key findings include Claude ranking as the second most-used weekly tool after Figma, and AI-assisted coding emerging as the top investment priority at 64%. The survey also identifies time as the primary barrier to adoption. This source is valuable for understanding AI adoption specifically within design practice.</p><h3>Reference</h3><p>Collins (n.d.). <em>Definition of workslop</em>. [online] <a href="http://collinsdictionary.com/">Collinsdictionary.com</a>. Available at: <a href="https://www.collinsdictionary.com/dictionary/english/workslop">https://www.collinsdictionary.com/dictionary/english/workslop</a> [Accessed 6 May 2026].</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=72cdc7834108" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>