Building a Better Geek

Watch here: https://youtu.be/zobiv1u9oJk Craig and Emmanuella debate whether AI-generated news is more trustworthy than traditional journalism—and what we lose if media dies. The Saturday Morning Shift Craig's Evolution: His Saturday mornings used to mean reading the Financial Review cover-to-cover. Now he starts there but ends up in Claude or Perplexity, asking all the questions journalists didn't answer. This raises an uncomfortable question: Should he trust AI-generated information more than journalistic organisations? The Argy-Bargy: This conversation stems from a Friday text exchange where Craig had a strong reaction to a podcast Emmanuella shared (though hunger from a failed drone food delivery may have contributed). His response sparked a debate about media bias, AI curation, and what happens to truth when traditional journalism collapses. The Media Bias Problem Emmanuella's Approach: She reads everything—left, right, moderate—because everyone has an agenda. Journalists have personal biases, plus organisational biases from their employers. She consumes The Australian, The Age, The Guardian, and AFR weekly, plus their podcasts whilst gardening or parenting. The Vienna Housing Example: Craig found an ABC article praising Vienna's rental market model (a post-WWI decision to treat housing as a right, not an investment). The article was overwhelmingly positive with no critical analysis. He wanted to know downsides, compare other markets, and challenge his comfort—but that information wasn't provided. Sins of Omission: Media doesn't just slant stories through what they say—they slant through what they don't say and which stories they don't cover. Craig found only one article on housing in a week, despite it being arguably Australia's top issue. AI as Research Tool Craig's Method: Uses Perplexity with multiple models (different AI providers have different biases) Asks for facts first, media interpretation second Can request exclusion of media sources to focus on official statistics Investigates topics like Australia's "gold plating era" (over-investment in energy infrastructure 10-20 years ago, still inflating bills today) The Insiders Frustration: Craig used to yell at Sunday morning political shows for asking inside-game questions instead of obvious, substantive ones. AI lets him ask his own questions and get deeper answers. The Critical Warning: Who Holds Power Accountable? Emmanuella's Concern: AI can only report what's been fed into it. Investigative journalists find information people don't want us to know—corruption, abuse, hidden agendas. Without them, who holds politicians, police, and corporations accountable? AI can't do that work. Not listicle writers ("5 Ways to Please Your Partner"), but real investigative journalists doing essential democratic work. The Business Model Crisis: Traditional media is economically challenged. New media (Substack, independent journalists) hasn't taken hold in Australia like elsewhere. Craig follows individual writers he trusts, not mastheads. The Trust Paradox Trust in traditional media: Historically low Trust in AI: No track record yet AI's Limitations: Nearly all training data is from the last 20 years (sparse pre-internet knowledge) Can pull data points from different sources measured differently, creating logical inconsistencies Very slanted by current trends and fads, lacking historical context Still has tech bro Silicon Valley bias in all major models The Power Shift Nobody's Noticing Craig's Key Insight: Traditional media's power is being challenged in ways not immediately apparent. If thousands of people seek understanding through their own prompts to different AI models, the mainstream media loses its gatekeeping function. Implications: Governments, political organisations, advertisers, and public health campaigns that relied on unified media channels now face diffuse, individualised information consumption. The message isn't controlled anymore. Key Takeaways Journalism remains essential: Garbage in, garbage out. AI needs high-quality investigative journalism to feed it—and society needs journalists to uncover what powerful people hide. The inflection point: Like the early internet when "tech people" adopted it before the world caught on, we're at a similar moment with AI-generated research. Every knowledge worker will soon be a "fast follower." Read widely, question everything: Whether consuming traditional media or AI-generated content, diversify sources and interrogate biases—including your own. Final thought: The line between genius and insanity often houses the early adopters. Both Craig and Emmanuella have been accused of being on both sides of that line.