AI with Kyle Daily Update 025

Today in AI: OpenAI Drops Bombshell Open Source Models + GPT-5 Coming Tomorrow?

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

Highlights

🎯 GPT-5 Released?

I’m writing this a few hours before the 6PM (London time) potential announcement of GPT-5.

All signs are pointing towards release Thursday evening. He tweeted it'll be "longer than usual, around an hour" because "we have a lot to show." Plus he posted that enigmatic Death Star image - classic Sam teasing at maximum level.

Kyle's take: I've cleared my entire evening to play with this. The Death Star image isn't just showing off and saying we’re going to destroy the competitors - I reckon it's actually generated by their new image model. Sam loves these multi-layered hints.

Still weird to align with The Empire…

What excites me most isn't whether it's "smarter" - they're all smart enough now. It's the unified interface.

Most of those 700+ million ChatGPT users have never touched o3 reasoning because it's buried in that confusing dropdown.

If GPT-5 combines everything into one model, suddenly everyone gets access to reasoning without having to think about it. That's the real game changer.

💥 Labour MP's AI Chatbot Goes Horribly Wrong

Remember that Labour MP Mark Sewards who created an AI avatar of himself? Yeah, it's going about as well as I predicted a few days ago.

The Express is running headlines calling him a "useless twit" after his chatbot refuses to answer any actual political questions. Ask about migration? "I'll pass that to my team." Ask about voting history? "Not available at this time."

Kyle's take: Two massive problems here. First, messaging failure - people think he's replaced himself with AI rather than supplemented his work. Second, anyone with legit questions to ask won’t get answers.

It's useless for anything controversial, which is literally what MPs are supposed to handle. I tried it myself and got kicked off the moment I asked about Gaza.

If you're not going to let your chatbot discuss politics, why make a political chatbot? He'll shut this down within a week is my guess. Fun experiment though.

🏗️ OpenAI's Open Source Strategy Becomes Clear

Yesterday's gtp-oss release makes perfect sense now. OpenAI dropped those 20b and 120b models knowing full well they're announcing GPT-5 today. It's classic product strategy - give away your current tech (or near enough) just before you obsolete it with something massive.

Kyle's take: This is great positioning. OpenAI just re-entered the open source fight exactly when Meta pulled back, keeping America competitive with China's open source push.

But more importantly for us entrepreneurs, these models could drive down inference costs significantly if you're building AI applications. The 120b model performance is close to o3 - at a fraction of the cost.

Why seemingly murder your top models? Well, if you are releasing something mega the next day (GPT-5) it’s a slick move.

🔥 Trump Fires Statistics Chief Over "Fake" Job Numbers

Donald Trump fired the Bureau of Labour Statistics commissioner after July job numbers showed only 73,000 new jobs created instead of the expected ~150,000. His reasoning? "I believe the numbers were phony, so I fired her." The same woman who was confirmed by a Republican Senate just last year.

Kyle's take: The irony is beautiful - he fires her for "political manipulation" whilst literally firing her for political reasons. You can’t make this up…

But the underlying trend is interesting. I don't think this is direct AI job replacement yet. Instead, it may be companies not creating new jobs because AI lets existing workers do more. Alongside other (many other!) economic factors of course - nothing is so neat and tidy.

We're seeing this in graduate schemes at big four firms and legal practices. It's not "fire and replace with AI" - it's "hire fewer people because AI amplifies productivity." That's the real threat to labour markets.

Member Question from Ronald: "Choosing between creative job and real job. Parents pushing towards nursing. AI affecting art jobs."

Kyle's response: Your parents aren't wrong about nursing honestly - it hits all three AI-proof criteria: non-algorithmic, empathy-based, and physical.

The creative industries were already difficult before AI, now they're becoming brutal. AI isn't coming for gallery art, it's coming for the commercial work that pays the bills - the everyday photography, marketing materials, fashion shoots.

That's where the money was, and that's what's disappearing.

My advice? Get a job that pays the bills in 3-4 days a week, then pursue creativity on the side without the pressure of it feeding you. Artists have been doing this forever - AI just makes it even more necessary.

This question was discussed at 24:30 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 and TikTok (follow and turn on notifications for Live Notification).

Audio Podcast on iTunes and Spotify.