<?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 Aneta Kula on Medium]]></title>
        <description><![CDATA[Stories by Aneta Kula on Medium]]></description>
        <link>https://medium.com/@poniatna?source=rss-db629950d59------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*fJ7_ydKO0V_Zn8Z0.jpg</url>
            <title>Stories by Aneta Kula on Medium</title>
            <link>https://medium.com/@poniatna?source=rss-db629950d59------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 05:54:25 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@poniatna/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[The chanterelles, the strawberries and the AI]]></title>
            <link>https://medium.com/@poniatna/the-chanterelles-the-strawberries-and-the-ai-5debe534b269?source=rss-db629950d59------2</link>
            <guid isPermaLink="false">https://medium.com/p/5debe534b269</guid>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Aneta Kula]]></dc:creator>
            <pubDate>Tue, 21 Oct 2025 20:06:36 GMT</pubDate>
            <atom:updated>2025-10-21T20:06:36.898Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*GMV43rTkWrvUkX3EgU4u7Q.png" /></figure><p>The idea for the app began where so many memorable meals do — with nostalgia. For travellers missing a dish discovered abroad, or immigrants longing for flavours from home, food often becomes a bridge to personal history. Across social media, countless posts reminisce about recipes that can no longer be properly recreated because the key ingredient is unavailable or prohibitively rare. From these everyday conversations emerged the spark: what if communities themselves could help one another rediscover those elusive tastes? A digital hub of culinary memory and collaboration.</p><p><strong>The core concept</strong></p><p>The app serves as a global recipe exchange centred on native dishes and the challenge of sourcing ingredients in new places. Users can upload family recipes that call for products unique to their homeland, while others within the community provide advice on where to find those ingredients locally, as well as suggestions for the best substitutes and where to purchase them. To make things more practical, the app’s regional feature helps identify restaurants that serve similar dishes, catering to those who prefer to celebrate nostalgia without the long hours in the kitchen.</p><p>An interactive, searchable map anchors the experience. It’s colour-coded to show cultural concentrations and pinpoints restaurants or stores specializing in regional cuisines. Participating restaurants can even share photos, recipes for signature dishes, and direct links to menus or delivery platforms. Tourists or newcomers unfamiliar with local neighbourhoods can view nearby attractions alongside dining options or do the reverse — search for an attraction and discover the foods from specific cultures available in its vicinity. The design is both intuitive and inclusive, meant to connect emotion, geography, and cuisine in one place.​</p><p>The app’s architecture is intentionally universal: every city hosts its pockets of culinary displacement and rediscovery. What began as an idea rooted in Toronto’s multicultural mosaic could easily scale worldwide. The model also opens multiple revenue channels — from advertising local markets and restaurants to featuring cultural festivals and small-batch food producers. In a moment of daydreaming ambition, the project’s creator even envisioned seeing it spotlighted on the Forbes innovation list. Only one obstacle stood in the way: limited coding skills and a lack of seed money. The solution seemed obvious — use AI. The urban legend goes that it is free, versatile, promising and user-friendly.</p><p><strong>Testing the promise of AI</strong></p><p>The initial assumption was straightforward: feed the concept to an AI model and it would generate a working prototype. However, the development soon showed that automation has its limits. The first step was personally selecting a recipe that represents home. Living in Toronto but with Polish roots, the author chose a Polish highlander tart made with kurki (chanterelle mushrooms) and bundz (a salty mountain cheese). Finding these ingredients locally highlighted the very need for the app: stores in the GTA mostly stock British, French, and Italian cheeses, and wild mushrooms are nearly impossible to find because foraging is not common practice, as it is considered dangerous and not without merit.</p><p>After securing the recipe, the next task was visualizing it. AI image generation promised an effortless solution — either describe the desired look or let the model interpret the recipe and generate a picture. The results were both entertaining and humbling. One generator, Freepik’s AI, produced a flawless image of a mouth-watering… strawberry tart with whipped cream.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*AaBt49FXExeTrxjO9mJ-8Q.png" /></figure><p>RunwayML offered something closer to reality but featured the wrong mushrooms and no cheese. Supplying reference photos improved the composition slightly, though the kurki appeared uncooked and oddly proportioned. Each outcome, while visually appealing, betrayed the model’s shallow grasp of culinary realism — a reminder that digital intelligence doesn’t equal understanding.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2eJaV7Qnnhl7wNVA_QgcCQ.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*AyiKgF-TYRxqNyd7u475aA.png" /></figure><p><strong>Mapping the mosaic</strong></p><p>Next came location mapping. Toronto, with its diversity, seemed the perfect testing ground. The author tasked Perplexity with generating a map of GTA regions colour-coded by cultural majority, expecting solid accuracy from the research-oriented AI. The result, however, resembled a fictional parallel of Toronto — similar, but subtly warped with an imaginary landscape, names, and borders.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*O6hPUf6iDiZr8_82O5oO3Q.jpeg" /></figure><p>ChatGPT performed better, offering an accurate map and even suggesting interactive features. Such contrasts revealed a truth common in AI workflows: consistency varies widely between models, even for structured requests.</p><p><strong>The deeper lesson</strong></p><p>The next phase — coding the interactive components — brought hesitation. While the author could handle culinary logic, photography, and design, programming felt like venturing into alien territory. Suddenly, very bad vibes came from vibe coding.</p><p>The author’s knowledge of coding is insufficient to identify and correct potential errors, and judging by the progress made so far, there will likely be numerous errors.</p><p>The experience reinforced a broader understanding of the promise versus the practice of technology. AI fosters the illusion of instant creation, but quality demands human skill, patient revision, and multidisciplinary knowledge. The models deliver surface polish — a convincing image, a functional map — but without contextual expertise, their results lack substance.</p><p>AI makes one think that everything is easy, automated, and ready when one wants it. But under the scrutiny of quality control, it turns out that what one gets may be something nice and appealing, but it is not accurate. Below those looks, there is no substance. To achieve that substance, one needs knowledge, experience, and creativity. AI is just a tool, like a hammer or a chisel, and it is no smarter than the person holding it. The outcome can be a bunch of scattered rocks or Venus from Milo.<br>For this reason, in a world full of AI-generated images, text, videos, and the like, having a well-rounded education will become increasingly important. As much as being an expert in one field is respect-worthy, it is going to be less desirable in a world full of AI-generated content, because as much as knowing one’s field is important, it is also important to know the difference between “kurki” and strawberries.</p><p>Ultimately, the experiment became an educational reflection on AI literacy itself. Craftsmanship still matters. In an era where algorithms blur authenticity, understanding how to distinguish truth from approximation may prove the most essential skill of all.</p><p>This assignment became a fun challenge for the author. The question of how far it can be taken without money and with the use of minimal skills seems to be an entertaining one. Please stay tuned, it will be continued.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5debe534b269" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>