Hey friend 👋 it’s Annie & Téa.
Top up your coffee ☕ — this week in AI is all about things getting faster, your home getting smarter, and rules getting clearer. Here’s the catch-up we’d share if you were sitting across the table.
📰 This Week in AI
1) OpenAI wants to make its own chips
Think of this like opening your own kitchen instead of renting one. If it works, AI tools could become quicker and maybe cheaper over time. Good news for everyone who uses them.
2) Google’s “Gemini” is coming to Home/Nest speakers
In plain English: your smart speaker is about to feel more like a chatty helper than a stiff robot. Better follow-up questions, more natural conversations, less “Sorry, I didn’t get that.”
3) New AI rules in Europe just kicked in
The EU now expects AI companies to be more open about how their models work and where content comes from. Translation: clearer labels and fewer mystery boxes for everyday users.
🔍 Demystifying AI
Why are chips a big deal?
Chips are the “engines” that make AI go. When a company builds its own, it’s like paving a private fast lane on the motorway: fewer traffic jams, smoother rides, and more control. For you, that can mean snappier responses and more reliable tools.
🛠 Tool of the Week – Readwise Reader
Too many tabs? Save articles, highlight the good bits, and let AI spit back a tidy summary so you actually remember what you read. Perfect for Sunday catch-ups or research sprints.
👉 Try it: readwise.io/read
💬 Prompt of the Week – For Testers / QA (no jargon!)
“Turn this feature description into a simple test checklist. Make two tables:
‘Works as expected’ tests.
‘Try to break it’ tests.
Each row should have: Short name, Steps to try, What we expect to happen.”
📈 What’s Trending
• Custom chips: Companies making their own = faster, possibly cheaper AI in the future.
• Gemini at home: Smart speakers that talk more naturally and remember context better.
• EU rules: More transparency so we know when and how AI is used.
• Open-weight models rising: Europe’s challengers are growing fast.
• FYI: ChatGPT had a brief outage this week; all good now.
🧩 Puzzle Corner (AI edition)
Last week’s answer: Footsteps.
This week’s puzzle: Spot the AI-written review
Three short reviews for a (fictional) electric kettle:
A) “Bought in April. Boils 1.7L in ~3m45s. Pour spout dribbles unless I tilt the lid ~45°. Keep-warm holds ~20 mins. Box came in generic ‘KitchenCo’ packaging, not the brand—odd but fine.”
B) “In today’s fast-paced world, this kettle is a game-changer. Its sleek design and unparalleled performance elevate every kitchen experience. I highly recommend it for all your boiling needs!”
C) “Light on the switch flickers once a week; doesn’t affect boiling. Limescale filter works. Replaced my old Russell Hobbs (took longer). Wish it were quieter.”
Question: Which one is most likely AI-generated — A, B, or C?
Hint (value add): Look for vague superlatives, reused phrases, and no specific, verifiable details. Humans tend to include messy, oddly specific stuff.
(Answer in next week’s issue 👀)
📅 Coming Next Week
• A fresh puzzle
• And 3 stories worth sharing over coffee
Stay curious 🪄
— Annie & Téa
🌟 Share the newsletter