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

Highlights

🚀 Elon's Open Source Gesture (That Nobody Needs)

Discussed at [03:40]

Elon Musk announced he's open sourcing Grok 2.5 and will release Grok 3 in six months (translation: two years in Elon-time). Sounds generous until you realise Grok 2.5 is pretty rubbish. Grok isn't even on the open source leaderboards. Meanwhile, Chinese models absolutely dominate the top spots - Qwen, DeepSeek, and others are running circles around everything else. It’s a fellow hollow gesture…but it’s a gesture at least.

Kyle's take: This feels like Musk having a dig at Sam Altman about OpenAI abandoning their open source roots. By continuing to (belatedly) release open source models Musk can claim the high ground.

But practically? Nobody's going to use these models. Grok 2.5 is inefficient, requires 500GB of VRAM, and gets destroyed by 24B models running on single GPUs. Plus if your company makes over a certain amount ($1m), you can't even use it commercially. It's better than nothing, but it's basically Elon proving a point whilst the serious open source innovation happens in China.

Source: Techcrunch

💔 Dating Costs Push Men Toward AI Girlfriends

Discussed at [17:18]

New study from an AI girlfriend app (biased sample, obviously) found that 43% of men say dating is too financially challenging, with nearly half skipping dates because they couldn't afford them. One-third would date an AI to cut costs, estimating it could save them hundreds per month. Some reckon it could save thousands.

Kyle's take: This is worrying but understandable. Sorta?

I met a bloke at my gym who spends £200 every week on dates - that's over £10k per year. When a pint in London costs nearly £10, you can see how a meal and drinks quickly hits £100-200. You can kinda see how that becomes an issue very fast.

Also remember that just because something seems odd right now that may just because it’s new to us. In the 90s, online dating was the butt of jokes, and within 10 years it became how most people meet.

We may be at a similar inflection point with AI companionship. Whether that's good or bad, I'll leave to your judgement.

Source: JOI app study of 1,000 users

Member Question: "What's the best way to ensure longevity as a B2B AI consultant and implementer?"

Kyle's response: It's all about your domain expertise. What most AI consultants and automation “experts” lack is deep domain experience in the industry they're working in.

You see loads of 20-year-olds selling automations to car detailing businesses they've never worked in - they don't have the industry knowledge to know what's actually valuable.

AI is a tool to solve business problems, so you need an intimate relationship with what those problems are. Combine what you've been doing in your non-AI life with your AI knowledge - it’s also how you outmanoeuvre big outfits like McKinsey or KPMG btw. They don't have that deep industry expertise.

Also, build an audience - distribution is becoming the 80/20. Become known as the manufacturing AI guy or the lady who teaches AI to educators. Carve out your niche and become top of mind in your particular industry. That will give you much more flexibility and longevity.

This question was discussed at [10:15] during the live session.

Member Question from User699944: "I'm fresh out of uni without domain expertise. How is it best to find a niche to specialise in?"

Kyle's response: Y Combinator calls it "going undercover" - go get a job with the express purpose of seeing what the core problems and difficulties are within that business, then step outside and solve them using AI.

You could build an internal tool for that company as your first customer. Or just work there for 3-6 months, figure out their problems, then leave and build the solution yourself. If you don't need immediate income, this is brilliant for fresh graduates.

Just be aware of IP ownership implications!

Alternative route is B2C - build an audience on Instagram, TikTok, Twitter and sell to individuals. B2B is better money (one customer paying £100k vs thousands at £20 each), but B2C might be easier without experience. Look at YoniMan - built his audience by posting daily for 500+ days, then launched successfully.

This question was discussed at [22:34] 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.