The skinny on what's happening in AI - straight from the previous live session:

Highlights

📊 BBC Study: "AI Wrong 45% of Time" - Missing the Real Story

Discussed at [00:00:59]

BBC-funded research claims AI assistants misrepresent news 45% of the time. But they tested free models without deep research modes. Oh and human journalists have 69% error rate according to 2020 study…

Kyle's take: This headline will spread everywhere - "AI wrong 50% of the time!" - but it's bollocks.

First, they used Gemini 2.5 Flash (the SPEED model, not accuracy) and GPT-4o without deep research or thinking modes. When I use Claude's deep thinking, I get 600 sources - they showed screenshots with TWO sources in their study.

Second, human journalists have 69% error rate according to Hungarian research, so AI's 45% is actually better.

Most importantly: in just six months, error rates dropped from 51% to 37%. In half a year. That's massive improvement. This is a snapshot using outdated methods - like testing a Ferrari in first gear.

AI does have problems. Big problems. And we need to be careful. BUT the studies are missing the bigger picture here and the fact that AI is getting better really fast.

The BBC has skin in this game - they're a news organisation worried people aren't reading anymore. In the next couple of years this will be a solved problem, but the "45%" will stick in people's minds.

Source: BBC / Ipsos study "News Integrity in AI Assistants" October 2025 / EBU Study

👶 1,500 Kids Worldwide Build AI Apps in Global Hackathon

Discussed at [00:18:12]

Brthrs Agency and Lovable ran worldwide kids AI coding hackathon - 80 locations, 1,500 children building apps with vibe coding tools. Kids built cool, creative projects with zero business constraints.

Kyle's take: This is brilliant - kids don't care about scaling, monetisation, or deployment strategies. The organisers shared some learnings (pictured above).

And they are items that we as adults could all stand to learn from!

The findings are gold: kids are genuinely supportive of each other (unlike competitive adults), they fail fast and iterate naturally (we need frameworks to teach us what kids do instinctively), and simple tools unlock massive talent.

This drives me mental about coding gatekeepers - when someone asks "should I use Lovable or Bolt?" dickheads on Reddit say "learn to code properly first." No! Lower the barrier, let people play, have fun FIRST. Once they realise they can build cool stuff, THEN they'll want to learn Python and deployment. Vibe coding is the gateway drug to building. Will you build enterprise SaaS immediately? No. But you'll have fun, and that's how everyone should start.

Source: Reddit post with Learnings

👓 Amazon's "Smart" Delivery Glasses: Surveillance in Disguise

Discussed at [00:34:46]

Amazon unveiled AI-powered delivery glasses with heads-up display for navigation, hazard detection, and proof-of-delivery photos. Fighter pilot tech for package delivery.

Kyle's take: They're selling this as "helping delivery associates navigate safely" - ha! This is 24/7 surveillance. Amazon will know if you stop for a cigarette, chat with someone, or take a bathroom break.

It's about logistics, cutting costs, increasing margins. That said, looks pretty cool! The green display looks straight out of Fallout 4 - appropriately dystopian…

I don't fault them for the business logic, but let's be honest about what this is: even MORE monitoring of staff who already can't take bathroom breaks. This tech will spread everywhere in the next few years. Arguably better than drivers looking at phones, but imagine having a map in your face all day. This is the future of work surveillance, dressed up as innovation.

Source: Amazon delivery innovation announcement

🏢 OpenAI Employee: "We're Not a Monoculture"

Discussed at [00:26:54]

OpenAI employee Aiden McLaughlin pushed back on conspiracy theories, explaining Sora and adult content weren't top-down decisions but came from passionate individuals fighting organisational headwinds.

Kyle's take: We tend to think everything's a master plan - Sam Altman sitting with accountants planning erotica features. Nah.

Sora was pushed by artists who love movie creation. Adult content came from a free-spirited professional athlete who believes in creative freedom. These are thousand-person companies with teams going off in different directions.

For instance I’m doing some work with Google’s teams. Right now they have about FIVE or SIX different vibe coding tools. They let teams build what they want, then see what bubbles up.

These corporations are HUGE. It's not all masterminded from the top - it's push and pull, bottom-up innovation meeting top-down strategy. Sam Altman isn't plotting every move with his fingers in every pie.

And that erotica thing? Honestly that one felt like a knee-jerk reaction to the Newsom bills that are forcing age-gating - basically a "fine, if you force age verification, we'll add horny content."

Member Question: "What's your thoughts on what a post-AGI world looks like?"

Kyle's response: Geoffrey Hinton says once we hit AGI, ASI follows shortly - the AGI takes over and works out the next steps itself. It won’t need us to help much.

AGI alone could be great - solving diseases, climate change. But ASI? We can't (by definition!) comprehend intelligence higher than us, like chickens looking at humans. They don’t understand what we are or what on earth we are doing. Humans haven’t encountered this phenomenon before.

So what does this mean for the future? The utopian version: general abundance, universal basic income. The (I think) realistic version: massive disruption, joblessness, poverty, war, famine. Humans are terrible at global cooperation.

Long-term I'm personally optimistic. It’s a necessary shift in the historical and technological sense. But there's an unpleasant period coming because we have no plans. We'll have a messy middle - could be decades or centuries. When people lose jobs, can't spend, capitalism itself breaks.

Remember that modern capitalism is only ~300 years old (ish) - we assume it's permanent as it is the paradigm we live in. But it's relatively new and might not survive what's coming.

This question was discussed at [00:13:24] during the live session.

Want the full unfiltered discussion? Join me tomorrow for the daily AI news live stream where we dig into the stories and you can ask questions directly.

Streaming on YouTube (with screen share) and TikTok (follow and turn on notifications for Live Notification).

Audio Podcast on iTunes and Spotify.