In this thrilling episode of Superintelligent, OpenAI’s GPT-5 arrived after two years of hype and fell short on everyday tasks. Side‑by‑side tests against GPT‑4.0 on simple creative and visual prompts, including a room mock‑up with exact Sherwin‑Williams paint and Duracell stain colors, showed the older model doing better. Both models even produced near‑identical poems about GPT‑5. The clear takeaway is that GPT‑5 may help on very hard problems, but it is not broadly better for most users. Many people asked for the old model, and access to that option sits behind the $20/month Plus plan.Monetization is reshaping how AI answers look and what sources appear. Current and planned revenue streams include ads inside answers, affiliate links, content licensing, and preferential ranking for paid partners. Licensed platforms surface more often, which is why Reddit links feel ever‑present, while X and Instagram block scraping and do not feed these tools the same way. Free tiers now face “shrinkflation,” where cheaper or smaller model variants serve more of those queries, pushing upgrades while cutting costs.Language around deals is muddy by design. “Partnership” often blurs with standard customer or API use, leaving users to assume deeper ties than exist. Apple’s strict policing of that word, even when years of joint engineering work occur, shows how brands manage perception and keep terms vague.AI browsers are the next big shock to the web. New tools like Dia and Perplexity’s Comet don’t just summarize pages; they act on them. In practice, Comet can open sites, click through flows, fetch data, and drop results into a new Google Sheet. Working inside logged‑in sessions, it can compile lists from social accounts without extra credentials. These browsers also make paywalls weaker by summarizing what’s behind them on demand. As people get used to “always be chatting” with pages, publishers face less click‑through and more extraction.Google’s AI Overviews push this trend further. The feature holds users on Google and reduces the visits and ad views that once funded reporting. The company says it still drives traffic to “forums, videos, blogs, and posts,” which skews toward social content. If newsrooms shrink, AI systems will end up remixing social posts over original reporting. That short‑term logic breaks the information economy that AI itself needs.Schools are moving fast to curb phones. A February survey found 77% of US public schools already restrict or ban phones. Entire states now enforce bell‑to‑bell bans, with Texas and New York joining for this school year. Many districts use Yondr pouches to lock devices during the day. Students complain about lunch and hall time without screens, but research points to higher grades, more focus, and more face‑to‑face talk. Parents cite emergency texting, including during shootings, yet schools have long managed urgent calls through the office and can set clear policies while phones remain on site.The broader culture shift is about attention. Fast feeds train people to kill boredom at once, raising the bar for stimulation so high that books feel dull. Bedtimes once nudged kids to read because nothing else was allowed. Simple home rules—like Faraday pouches at night—restore sleep and quiet. The push to “legalize boredom” is not nostalgia; it is a practical fix for concentration, curiosity, and social skills. Outside the US, contrasting feeds underline the stakes: ByteDance runs a very different domestic product in China that pushes duty and learning, while the global TikTok app leans into sticky short clips.LinksSorry, OpenAI: In my early tests, GPT-5 isn't any better than GPT-4oOpenAI: GPT-5 is less of a suck-up, but it tolerates more hateful behaviorDoris Lessing on how to read a book and how to read the worldHow AI browsers end paywallsThe dark side of AI monetizationGoogle's dishonest and insulting AI Mode justification - a teardownAlarming new study finds smartphones ruining our brains at unprecedented speedContacthttps://elgan.com/abouthttps://
[email protected]:00 The AI Moment: Introduction to GPT-502:54 The Disappointment of GPT-505:53 Monetization Strategies in AI08:42 Content Licensing and AI Bias11:40 The Nature of Partnerships in AI14:47 The Cult of Apple and AI Development17:38 The Shrinking Quality of Free AI Models20:45 Exploring Kagi and Model Selection23:50 The Future of AI: Delusions and Realities24:43 Leadership and Product Launch Failures27:33 The Petty Takedown of Sam Altman28:33 The Origins of the Podcast33:57 Exploring AI Browsers43:11 The Future of Browsers and Google's Monopoly46:16 Monetization of Information52:14 The Impact of Cell Phone Bans in Schools01:05:28 The Future of Technology and Reading Habits This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.superintelligentpodcast.com
Australia