AI/Tech Angle A, June - Secondary
Claude vs Gemini. GPT-7 vs Llama 5. Which AI lab ships AGI first. These are live Kalshi markets with real money on both sides, updated in real time as releases land. The person who follows model cards and tracks evals has a genuine edge here. If that's you, trade it.
TODAY IN AI
3 things that happened while you were busy
1. Aside ships an AI browser that logs into your accounts and gets tasks done.
The agentic browser turns your browsing history into on-device memory, then uses autofill to sign into your accounts and finish tasks without any hand holding. The startup, backed by Y Combinator, claims the top spot on three browser-agent benchmarks, and its launch demo shows the AI hunting down and canceling unused subscriptions. Try Aside yourself or watch the launch video that passed 1M views.
2. OpenAI updates GPT-5.5 Instant and reveals its first in-house chip.
The refreshed GPT-5.5 Instant is rolling out now, tuned to read the intent behind your prompts so it feels more conversational. OpenAI also unveiled Jalapeño, its first custom chip, co-developed with Broadcom to run inference and trim its dependence on Nvidia. Learn how the new chip works.
3. Google is about to lose two more of its top Gemini researchers.
Bloomberg reports that Jonas Adler and Alexander Pritzel, both key contributors to the Gemini model, are preparing to leave for Anthropic. The exits would extend a steady talent drain at Google, which recently lost Nobel laureate John Jumper to Anthropic and saw veteran researcher Noam Shazeer leave for OpenAI.
FROM THE FRONTIER

Made with Midjourney
Europe has five years to fix its AI gap, or it gets left behind for good.
Europe 2031. Eight researchers just published Europe 2031, a thought piece calling for the most ambitious political agenda in post-war European history. Their core argument is blunt: the bloc has fallen behind in AI and the window to catch up is closing. Without action, they warn, Europe loses the power to shape its own future within five years, not within a decade.
Sliding into irrelevance. The essay says European leaders keep misreading the speed of the race, meeting it with skepticism instead of urgency. The result is stark. The bloc has few homegrown frontier models and controls roughly five percent of global compute, against America's 80 percent. With almost no real influence left, the authors sketch out how the next few years could unfold.
Looking ahead. Their 2031 scenario is grim. The US owns the cognitive frontier, China leads in robotics, and everyone else negotiates from weakness. Europe, they argue, would be forced to pick a side or, in their words, wither away in isolation.
Righting the ship. The fix they propose is not subtle. Pour money into compute, rally a coalition of other lagging nations to build bargaining power, and bet hard on winning the physical-AI wave. The piece rhymes with Citrini Research's The 2028 Global Intelligence Crisis, which runs a similar thought experiment for the American economy. The uncomfortable part is the price tag, which may be larger than Europe has ever been willing to pay.
IN THE KNOW

What people are actually watching and sharing
Claude Tag. Andrej Karpathy is calling Claude Tag a new paradigm for working with Claude, though plenty of people push back and say the update is nothing special. His full comment and the debate around it have drawn 4.5M views.
Consent Registry. Want to stop AI tools from using your likeness? You can add your name to the Human Consent Registry, a new public record for stating that preference.
Back in time. AI has spawned a new content genre of historical point-of-view videos. One creator walks you through a day in the life of 1536 London, and it has racked up 5M views.
2000s throwback. Someone reimagined Claude, ChatGPT, Gemini, and Grok as CD cover art from the early 2000s, and the results are oddly perfect.
Dead internet. A Redditor found a World of Warcraft server with no real humans on it, just 1,800 DeepSeek bots going about their day. The poster calls it the dead internet theory in action, though many commenters disagree (5.8K upvotes).
PROMPT STATION
Turn any TV show into a set of collectible fridge magnets
Here is one for the binge-watchers. Paste it into ChatGPT's image generator with any show or movie, and it returns a flat-lay of nine hyper-real resin fridge magnets, chibi versions of the main characters up top and the show's most iconic props below. It is a quick way to turn a favorite series into shareable fan art in about a minute.
Suggest 10 high-impact AI automations I can set up this week that together save 5+ hours. My tools are [TOOLS] and my main bottlenecks are [PAIN POINTS]. For each automation, include the exact tools used, a clear description of what it does, estimated setup time, and estimated hours saved per week. Also add a difficulty level (low/medium/high) and a brief ROI breakdown showing total weekly time saved across all automations. Focus only on practical, implementable workflows with immediate impact.A flat-lay collection of exactly 9 hyper-realistic 3D resin fridge magnets arranged on a bright pure white surface in a clean 3x3 grid, themed around [SHOW/MOVIE]. At the very top center, a wide rectangular text magnet spells "SHOW/MOVIE" in bold 3D block letters styled in the show's exact signature typography and colors. Below it, 9 magnets in a clean 3x3 grid: the first row contains 3 chibi full-body character figurines of the most iconic characters, perfectly detailed and instantly recognizable. The second and third rows contain 6 of the most iconic props, locations, and symbols from the show, objects any casual viewer would immediately recognize. Each magnet is small, neatly spaced with generous white space between them, slightly raised with a soft drop shadow beneath. Bright white background, warm studio lighting from above, photorealistic product render, top-down flat-lay angle, 2:3 aspect ratio.


