<?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 Ave Shalom on Medium]]></title>
        <description><![CDATA[Stories by Ave Shalom on Medium]]></description>
        <link>https://medium.com/@aveshalom?source=rss-feb7a7e4fb26------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*3ULW91laPX3IpYIKWsFuew.jpeg</url>
            <title>Stories by Ave Shalom on Medium</title>
            <link>https://medium.com/@aveshalom?source=rss-feb7a7e4fb26------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 09:49:15 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@aveshalom/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[Paving the Desire Lines: Leading Design in the Age of AI]]></title>
            <link>https://medium.com/@aveshalom/paving-the-desire-lines-leading-design-in-the-age-of-ai-105eb0580def?source=rss-feb7a7e4fb26------2</link>
            <guid isPermaLink="false">https://medium.com/p/105eb0580def</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[design-leadership]]></category>
            <category><![CDATA[product-design]]></category>
            <category><![CDATA[ux-design]]></category>
            <category><![CDATA[sustainability]]></category>
            <dc:creator><![CDATA[Ave Shalom]]></dc:creator>
            <pubDate>Fri, 13 Nov 2026 17:23:04 GMT</pubDate>
            <atom:updated>2026-11-13T18:01:29.242Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="Picture of a person walking a dirt path next to a paved park path." src="https://cdn-images-1.medium.com/max/1024/1*3zD5JdMNOKbPAyTX1MhVYw.png" /><figcaption>AI generated</figcaption></figure><h3>Off the paved path</h3><p>Frederick Law Olmsted designed some of the most beautiful parks in the United States. Rose gardens, meadows, ponds, brambles, museums, all stitched together by carefully planned paved paths.</p><p>And yet people still step off them.</p><p>Walk through any park and you’ll see the worn ribbons of dirt cutting across the grass. Urban planners call them desire lines. They’re the routes people actually take, as opposed to the ones someone drew for them. Nobody coordinates them. Enough people want the same shortcut, and the grass gives way.</p><figure><img alt="Illustration of Prospect Park, Brooklyn, the design is considered a classic example of Olmstedian design, 1871. Olmsted Archives" src="https://cdn-images-1.medium.com/max/650/0*zdnOKiW7H12hgC-s" /><figcaption>Prospect Park, Brooklyn, considered a classic example of Olmstedian design, 1871. Olmsted Archives</figcaption></figure><p>I’ve been thinking about desire lines a lot lately, because they help explain my career. And I think they explain what’s happening to design right now with the introduction of AI.</p><h3>Growing up before the paths</h3><p>I started in tech before user experience design had a defined method. There were no bootcamps and no curriculum. There was no playbook. Mostly there was intuition, and a lot of walking across the grass.</p><p>I remember doing it all at once: user research, market analysis, sketching journeys, and building coded prototypes so we could feel how an interaction worked before we built it. Nobody called that UX yet. My titles traced the same path: Graphic Designer, Interactive Designer, Creative Director, Experience Design Director. And now… we’ll see.</p><p>Over time, the methods I used by instinct became the established practice of today. That wasn’t because of me. It was because of the collective. Enough of us were cutting the same paths across the same fields, and eventually someone paved them. Sociologists have studied this for decades. Everett Rogers’ work on the diffusion of innovations shows that a practice spreads when it’s clearly better, fits how people already work, and is easy to try and see. Practices that don’t meet that bar stay with an individual or a small group. Practices that do, become careers.</p><p>There’s a logic to it. People find the easier way, and then the easier way becomes the norm.</p><h3>Fresh grass</h3><p>Here we are again. All of us are starting our careers over with AI. We don’t know the direction, and we don’t have the playbooks. The mood is a mix of excitement, fear, worry about the planet, overconfidence, and nervous confusion. Some days I feel all of them before lunch.</p><p>But the first trails are already being worn in. We have spec frameworks. We’re building our own skills, agents, harnesses, and rules, and learning where we want the AI to be predictable and where we want it to surprise us. Policies for responsible use are being written. So are policies for where data centers get built. Slowly, we’re getting comfortable with our new AI companions.</p><p><strong>Nobody planned these paths. We’re making them by walking.</strong></p><h3>Deciding what to pave</h3><p>I lead a team of Experience Designers. Or rather, Agentic Designers. Or maybe Design Engineers. Even the job title is still a desire line.</p><p>I see two patterns on my team: new curiosity and rebellion. I understand the rebellion. I grew up in the arts, and I miss the tactile feel of charcoal on paper. I even miss pushing pixels in Figma, which is funny, because that was once the abstraction we were mourning. Now the abstraction moves again. We write in natural language. We spend more of our time setting up guardrails, product requirements, and memory systems than drawing. Only then do we ask AI to synthesize the research or build a UI from our requirements and design-system markdown.</p><p>There’s a tug to rebel. For me, there’s a stronger tug to experiment and imagine what might be possible.</p><p>Leading in this moment feels like we’re all mentoring each other. The old mentor-mentee ladder has flattened into something more like shared discovery. Learning researchers Jean Lave and Etienne Wenger would recognize it as a community of practice, where people learn by doing the work together rather than by being taught.</p><p>But flat doesn’t mean formless. My team still needs structure and operating models to grow, deliver, and thrive. <strong>That’s the leader’s job now: deciding which desire lines to pave so people can move fast, and protecting enough open grass for new ones to form.</strong></p><p>Here’s one we paved. A couple of my colleagues built a “second brain” for consultants starting a project. It builds shared memory over time, keeps deliverables on track, and keeps everyone on the same page. It’s robust, and honestly, beautiful. Here’s one we left open: my team and I are experimenting with layered versions for UX, a shared foundation with design-specific layers on top stored in one or more repos. We don’t know yet where it should live. That’s the point.</p><h3>Make your own line</h3><p>The wisdom of experience still holds. Critical thinking, craft, and a curator’s mindset matter more now, not less.</p><p>The research backs this up. In a 2024 study in <em>Science Advances</em>, Anil Doshi and Oliver Hauser found that AI-generated ideas made individual stories more creative, especially for writers who scored lower on a baseline creativity test. But the AI-assisted stories also looked more like one another. <strong>AI raises the floor and narrows the range. If everyone follows the same suggestion, everyone ends up on the same paved path.</strong></p><p>A 2025 survey of 319 knowledge workers by researchers at Carnegie Mellon and Microsoft found something similar about thinking. The more people trusted the AI, the less critical thinking they did. The more they trusted <strong><em>themselves</em></strong>, the more they applied critical thinking. The work itself shifted toward verifying, integrating, and stewarding what the AI produces. That’s curation, and it’s a skill you can teach.</p><p>So this is what I tell emerging talent. Push past the slop. Set the bar high, and make the tools execute your vision. Have a vision. Imagine what might be possible. Be curious, experiment, and above all, be thoughtful.</p><p>Token waste is the absence of those things. It has a cost beyond quality, too. Every prompt runs on electricity somewhere. Sasha Luccioni and colleagues found that energy per use can vary by orders of magnitude depending on the model and the task, and that generative tasks cost far more than simple ones. A careless prompt loop isn’t just lazy design. It burns real energy for nothing.</p><p><strong>Our prompt histories may be the new portfolio. They show how we think, and how much we waste.</strong></p><h3>Illogically logical</h3><p>People bring patterns to design. We also bring anti-patterns. Biologically, we’re always hunting for the easier way: the loophole that gets us what we want faster, cheaper, and with fewer calories burned. Cognitive scientists Evan Risko and Sam Gilbert call one version of this cognitive offloading. We hand off mental work to tools when we believe the tool will do it better, even when that belief is wrong.</p><p>That’s what desire lines are. <strong>And here’s the part I love:</strong> <strong>once a desire line gets paved, it opens new destinations, and new desire lines start forming off of it. The pattern keeps moving. </strong>It’s hard for a trained logic system to predict, because it keeps breaking its own rules.</p><p><strong>We are illogically logical. We’re strangely compelled to make what nobody expected.</strong></p><p>In 1967, a young art student named Richard Long stepped off the path on purpose. He walked back and forth across a field in Wiltshire, England, until the grass gave way, then photographed what was left. He called it <a href="https://www.tate.org.uk/art/artworks/long-a-line-made-by-walking-p07149">A Line Made by Walking</a>. It isn’t a shortcut, and it doesn’t lead anywhere. Long has described the idea as making a footpath going nowhere. It’s a desire line whose only destination is the desire to make something.</p><p>Another example is Antoni Gaudí’s Sagrada Família reached a milestone more than a century in the making. The design was his. Construction began in 1882, and he led it until his death in 1926, when only a small part had been built. Most of his original plans were lost during the Spanish Civil War. Generations of architects, engineers, and artisans rebuilt his intent from surviving models and fragments, eventually with the help of computational tools. In 2026, the centenary of his death, the cross was placed on the central Tower of Jesus Christ and the exterior was finished. The interior is still underway.</p><figure><img alt="Sagrada Familia Basilica interior by architect Antoni Gaudi" src="https://cdn-images-1.medium.com/max/1024/1*thbnKIFYUgdsrUp4JolQ1Q.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@star7a?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Edwin Chen</a> on <a href="https://unsplash.com/photos/gold-and-white-chandelier-turned-on-in-a-room-VCr89JhUOlQ?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></figcaption></figure><p><strong>It’s as impractical as it is beautiful. </strong>No optimizer would have proposed it. But it also shows where I think AI belongs. Gaudí’s vision didn’t come from a machine, and it didn’t need to. Machines and people helped carry a human idea further than one human could.</p><p>AI learns from us, and increasingly from itself. But what it makes is still a reflection of what it predicts a person would do. <strong>It doesn’t step off the path because it’s curious where the grass leads. That part is still ours.</strong></p><h3>Back to the park</h3><p>I’m not worried about what we’ve set in motion. I might be wrong about some of what I’ve said here. What matters more to me is how we treat each other and the planet while we figure it out.</p><p>That means building structures that give all of us a safe place to coexist: people, tools, and the ground everything runs on.<strong> </strong>It means counting the energy behind every prompt as a design cost, not someone else’s problem. <strong>And as a leader, it means helping my designers grow into inspired and inspiring people. Stewards of a better tomorrow.</strong></p><p>Olmsted drew the paths. People walked their own. The best parks learned from both.</p><p><strong>That’s the kind of leader I want to be. Pave what helps people move, and leave room for the grass.</strong></p><h3>References</h3><ul><li>Doshi, A. R., &amp; Hauser, O. P. (2024). <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC11244532">Generative AI enhances individual creativity but reduces the collective diversity of novel content</a>. <em>Science Advances</em>, 10(28). doi:10.1126/sciadv.adn5290</li><li>Lee, H.-P., Sarkar, A., Tankelevitch, L., et al. (2025). <a href="https://www.microsoft.com/en-us/research/?p=1135061">The Impact of Generative AI on Critical Thinking</a>. <em>CHI </em>’25. doi:10.1145/3706598.3713778</li><li>Luccioni, A. S., Jernite, Y., &amp; Strubell, E. (2024). <a href="https://arxiv.org/abs/2311.16863">Power Hungry Processing: Watts Driving the Cost of AI Deployment?</a> <em>ACM FAccT </em>’24. doi:10.1145/3630106.3658542</li><li>Risko, E. F., &amp; Gilbert, S. J. (2016). <a href="https://discovery.ucl.ac.uk/1508770/">Cognitive Offloading</a>. <em>Trends in Cognitive Sciences</em>, 20(9), 676–688.</li><li>Foster, A., &amp; Newell, J. P. (2019). <a href="https://ilstu.academia.edu/afoster/Papers">Detroit’s lines of desire: Footpaths and vacant land in the Motor City</a>. <em>Landscape and Urban Planning</em>.</li><li>Rogers, E. M. (1962). <em>Diffusion of Innovations</em>. Free Press.</li><li>Lave, J., &amp; Wenger, E. (1991). <em>Situated Learning: Legitimate Peripheral Participation</em>. Cambridge University Press.</li><li>Sagrada Família history and 2026 milestone: <a href="https://www.dezeen.com/?p=2050805">Dezeen</a>, <a href="https://www.domusweb.it/en/news/gallery/2026/03/02/sagrada-familia-completed-world-tallest-church.amp.html">Domus</a>.</li><li>Long, R. (1967). <a href="https://www.tate.org.uk/art/artworks/long-a-line-made-by-walking-p07149"><em>A Line Made by Walking</em></a>. Tate, London. © Richard Long, DACS.</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=105eb0580def" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>