<?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 Ignacio Amaya on Medium]]></title>
        <description><![CDATA[Stories by Ignacio Amaya on Medium]]></description>
        <link>https://medium.com/@xinofekuator?source=rss-654aa59fb446------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*7ZOjVJNIgfxGh7MwcfqdRA.jpeg</url>
            <title>Stories by Ignacio Amaya on Medium</title>
            <link>https://medium.com/@xinofekuator?source=rss-654aa59fb446------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 10:58:51 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@xinofekuator/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[Why My AI’s Ice Cream Advice Is a Warning About Where We’re Headed]]></title>
            <link>https://medium.com/@xinofekuator/why-my-ais-ice-cream-advice-is-a-warning-about-where-we-re-headed-a1f3c96c9984?source=rss-654aa59fb446------2</link>
            <guid isPermaLink="false">https://medium.com/p/a1f3c96c9984</guid>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[society]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[personal-essay]]></category>
            <category><![CDATA[ethics]]></category>
            <dc:creator><![CDATA[Ignacio Amaya]]></dc:creator>
            <pubDate>Sun, 19 Oct 2025 15:32:00 GMT</pubDate>
            <atom:updated>2025-10-19T15:32:00.686Z</atom:updated>
            <content:encoded><![CDATA[<h3>Why My AI’s Ice Cream Advice Got Me Thinking About the Future</h3><p>I was down with a cold last week but still had to take our dog out for walks while my wife was at work. It’s been freezing here in Finland — not exactly ideal recovery weather. During one of those walks, my wife sent me a YouTube video claiming lemon ice cream could help a sore throat.</p><p>I had vanilla ice cream in the freezer, and the thought of it with some leftover warm mud cake sounded amazing after days of eating nothing but soup. Still, ice cream while sick felt odd, so I didn’t trust the video right away.</p><p>So, while carrying my dog’s poop bag in one hand, I asked Gemini on my phone by voice: “Is ice cream good for a sore throat with the flu — and is vanilla or lemon better?” The phone answered almost instantly, saying vanilla was better — the acidity in lemon could make my throat feel worse.</p><figure><img alt="Photo of Nina, the author’s dog, during a chilly autumn walk in Finland — the moment that led to asking an AI for advice while sick." src="https://cdn-images-1.medium.com/max/1024/1*gLsV0IwwO7soTErFMCZXSQ.jpeg" /><figcaption><em>Nina, our dog, during that freezing walk — the moment that started this whole story.</em></figcaption></figure><h3>🍦 How Convenience Builds Trust</h3><p>Was that health advice itself a real risk? Maybe. It’s replacing what should be a quick check with a doctor — or at least a trustworthy source. But the greater danger isn’t the accuracy of the answer; it’s how my reliance and trust will grow over time.</p><p>Right now, I use these tools like I’m talking to a friend I don’t fully trust. I wouldn’t rely on them for major life decisions, but for small, low-stakes questions, they’re a huge time-saver.</p><p>The alternative? Endless Google results — or worse, scrolling through Reddit threads or the YouTube comments, many of which were contradictory or raised reasonable but unchecked risks, like the danger of eating ice cream with pneumonia. The AI was fast, specific, and sounded personal, which made its answer feel tailored to me.</p><p>So, after the dog walk, I had my mud cake with vanilla ice cream and actually felt a bit less throat pain. (Placebo effect? Maybe.)</p><figure><img alt="Photo of a slice of chocolate mud cake served with vanilla ice cream, illustrating the author’s comfort food moment after asking an AI for advice while sick." src="https://cdn-images-1.medium.com/max/1024/1*FSa_TURXVZ-fJ0VllNFLhw.jpeg" /><figcaption><em>Mud cake and vanilla ice cream — comfort food for a sick day.</em></figcaption></figure><h3>🤖 <em>We’ve Seen This Movie Before</em></h3><p>That small moment — asking about ice cream while holding my dog’s leash — seems harmless. But it’s exactly in those small, everyday interactions where trust starts to build, and where the line between helpful assistant and subtle influencer can quietly blur.</p><p>This speed and convenience feel a bit dangerous. Without proper regulatory guardrails, we could be creating a direct, frictionless channel for corporate manipulation.</p><p>Imagine a future where the AI’s answers aren’t based on medical wisdom, but on commercial incentives. The danger isn’t which flavor to choose, but whether the AI recommends eating ice cream for your sore throat because an ice cream company paid to boost that suggestion.</p><p>Generative AI still lacks a sustainable business model. And if history is any guide — think social networks — the optimization target will inevitably shift to engagement and conversion. That’s when we start crossing into dangerous territory.</p><p>We’ve already seen what happens when regulation lags behind innovation. Social media platforms became breeding grounds for polarization, misinformation, and mental health issues. As a society, we’ve largely failed to contain those side effects — and, frankly, the outlook for doing better with AI isn’t very encouraging either.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*J-85JSzNNr9ChTFe" /><figcaption>Photo by <a href="https://unsplash.com/@maria_shalabaieva?utm_source=medium&amp;utm_medium=referral">Mariia Shalabaieva</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h3>🛠️ Building the Guardrails We Missed Last Time</h3><p>With personal AI companions, the risks are amplified — because it’s a voice in your pocket, tailored to you. AI will grow more intimate — shaping what we eat, buy, and even believe — and more controlled, as corporate incentives quietly turn helpful assistants into persuasive salespeople.</p><p>So what should we do about it?</p><p>I’m not sure we have the answers yet, but if we’ve learned anything from social media, it’s that waiting until things go wrong isn’t a great strategy. We need at least some early guardrails — not perfect ones, just enough to slow down the worst incentives.</p><p>Maybe that starts with ideas like:</p><ul><li><strong>Transparency: </strong>We should know where advice comes from and who benefits from it. Open projects like Hugging Face’s model cards or EleutherAI show it’s possible to make models more understandable — revealing what data they use and what biases they carry. And open datasets like LAION highlight why that visibility matters: they remind us how much of the internet’s messiness ends up inside these systems. Transparency shouldn’t be optional — it should be built into how these tools are created and deployed.</li><li><strong>Accountability:</strong> Some companies like to showcase their investments in fairness and safety — but to me, many of these efforts still feel like PR shields rather than real commitments. True accountability shouldn’t rely only on good intentions; it needs clear incentives and independent oversight. How to do that without killing innovation is the tricky part — and honestly, I don’t have a perfect answer. But whatever rules we build should help AI grow in directions that benefit society, not just quarterly earnings.</li></ul><p>Because if the past decade taught us anything, it’s that unregulated systems rarely optimize for our well-being — only for our attention.</p><p>And that brings me back to that small moment with the ice cream.</p><p><strong>If an AI can confidently suggest what to eat while I’m sick, what happens when the questions get bigger?</strong> We might not notice how easily convenience turns into quiet influence — one helpful answer at a time.</p><figure><img alt="A realistic image of a vanilla ice cream cone melting into a circuit board, representing how ordinary human choices can blend with the hidden influence of AI systems." src="https://cdn-images-1.medium.com/max/1024/1*yeoNM8ZkHNkjRUb07PHw7w.png" /><figcaption>AI-generated image: vanilla ice cream melting into a circuit board.</figcaption></figure><p><em>Note: I started this post by recording my ideas out loud, then used AI tools to help shape, edit, and refine the writing through several drafts.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a1f3c96c9984" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>