Daily Briefing — May 21, 2026

Samsung's union and management walked out of their latest round of talks on Wednesday without a deal, and now more than 70,000 workers are heading into an 18-day strike. The union says Samsung has been pocketing AI driven profits without passing any meaningful share to employees. Management says the union is overreaching, especially on behalf of workers at units that are actually losing money. Both sides are pointing fingers. Nobody blinked.

This is bigger than a typical labour dispute. Samsung and SK Hynix together produce roughly two thirds of the world's memory chips. Memory chips are not a niche component. They are in your phone, your laptop, the data centre infrastructure running every AI tool your company is probably already paying for. A prolonged stoppage at Samsung does not stay in South Korea. It ripples through every supply chain that touches semiconductors, which at this point is basically all of them.

The South Korean government is so nervous about this that officials are floating emergency powers to force a settlement, which tells you everything about the stakes. When governments start reaching for tools they almost never use, the situation is serious.

AI's profit surge is also widening the gap between the companies cashing in and the workers running the operations. That fight isn't going away.

Google just made it possible to build a native Android app and push it to your phone in minutes, no coding background required. At Google I/O, the company updated its AI Studio tool to support vibe coding for mobile, meaning you describe what you want, the AI builds it, and you export it directly to your device. Right now it is scoped to personal utility apps, and the Play Store rules still apply if you want to go public with anything, but the ceiling on what a non-developer can ship just got a lot higher.

Vibe coding started as a desktop novelty — people automating spreadsheets or building little tools for themselves. Mobile changes the stakes, because the phone is where you actually live. Building something that fits exactly how you think, solves the problem you have, and sits in your pocket is a different proposition from anything we've had before.

The gap between "person with a good idea" and "person who can ship a working product" is closing fast. That changes what skills matter, what gets you noticed, and honestly what excuses are left for not building that thing you have been talking about for two years.

Michael Wooldridge has been working in AI for more than 30 years. He's an Oxford professor with 500-plus scientific papers and ten books to his name, and he's one of the rare people who can explain this stuff without either hyping it into oblivion or dismissing it as a fad. His take on the robot takeover? He's not worried about it. The dangers he cares about are more mundane and more here-and-now: how Silicon Valley's entrepreneurs keep misusing the tools they build, often in ways game theory could predict if anyone bothered to look.

He's enthusiastic about technology even while staying clear-eyed about who is steering it. He brought a robotic dog into his Royal Institution Christmas lectures to show how the gap between "looks intelligent" and "actually is intelligent" is wider than the headlines suggest. Fear of an all-powerful AI is a convenient distraction from the boring stuff already shaping our lives: algorithmic feeds, opaque models deciding who gets a loan or a job, AI tools deployed at work without anyone asking the people doing the work.

The loudest people in the AI conversation are usually the ones with something to sell. A 30-year veteran who isn't running a startup and isn't trying to raise a fund is a useful counterweight.