Fake Menus and AI Friends: The High Cost of Easy Answers

Welcome to The Drafts, a podcast about building a business while working full-time with honesty, no highlight reels.In this episode, I'm talking about AI, but not in the way you might expect. It started with a menu I couldn't read at a Japanese restaurant. The images were AI-generated, glossy, perfect, and completely fake. And once I started noticing, I couldn't stop. What are we actually solving for when we implement AI everywhere? And who pays the price?You'll hear:Why restaurants use AI-generated images of fake food instead of photographing the real thingThe uncanny valley we're living in: AI images good enough to pass at a glance, wrong enough to make you feel lied toWhat LLMs actually do (they aggregate, they don't create) and why the Detroit: Become Human painting scene captures it perfectlyReal examples of AI failures: law firms submitting hallucinated case citations, a consultancy caught twice using LLMs for government reports, a chatbot selling an $80,000 car for $1Why workplaces implement AI not because it works, but to say they're doing itThe hidden costs: data centers consuming electricity like small countries, millions of gallons of water daily, carbon emissions equivalent to hundreds of flightsThe labor cost: Kenyan moderators paid $2/hour to read graphic content with little psychological supportThe artist cost: billions of images scraped without permission to train models that undercut the people who created themThe linguistic divide: English speakers get the best AI, everyone else gets leftovers and the gap compoundsThe loneliness epidemic by the numbers (21% of U.S. adults feel lonely, 30% of young adults feel lonely daily, 73% cite technology as a contributor)Why AI companions create manufactured attachment, not real connection and why solving loneliness with chatbots won't workWhat would actually help: third spaces, community infrastructure, asking friends for help instead of buying solutionsThe legitimate uses of AI: language learning (with limitations), assistive technology, using it as scaffolding (not the final output)If you've ever felt uneasy about AI without being able to articulate why, or wondered whether the convenience is worth the trade-offs, this is for you.The "right" use of AI isn't the one that sounds impressive in a boardroom. It's the one that actually solves a problem without creating worse ones.Full show notes and resources are available at thedraftspodcast.substack.comMusic in this episode:"The Drafts Theme" by Madhumita PrasadDisclaimer: The content of this episode is for educational and informational purposes only and represents the personal opinions of the host. All references to specific companies, events, or corporate practices are based on publicly available news reports, court documents, and cited publications. These citations are provided for critical analysis and commentary on matters of public interest. The views expressed here are solely those of the host and do not reflect the views or policies of any current or former employers, clients, or affiliated organizations.