Silicon Valley Is Building the Future While Gen Z Watches It Burn
AI companies are racing ahead on capabilities nobody asked for, regulators are finally acting, and young people are checking out. Here's what breaks next.
The disconnect is stunning. OpenAI is pitching four-day work weeks as a feature of the AI future. Meta just shipped a new superintelligence model. Anthropic is hoarding a cybersecurity breakthrough because it’s too dangerous to release. And half of Gen Z—the people supposed to inherit all this—are souring on the whole thing.
This is what happens when innovation outpaces either trust or consensus about whether we actually want it.
The Competence Trap
Here’s what’s actually happening underneath the hype cycle: the AI industry has moved from “can we?” to “we will, whether you’re ready or not.” Meta’s new Muse Spark model performs better than their previous systems. Anthropic built something so capable at finding vulnerabilities they won’t let it out. OpenAI’s suggesting bosses give people an extra day off because AI will handle the overflow. It’s all technically impressive and socially tone-deaf.
The competence trap works like this. Companies pour billions into R&D. They make breakthroughs. They announce them. The narrative becomes “this is inevitable, so you’d better adapt.” But nobody actually asked if we wanted AI systems that could be weaponized against networks, or if we needed to solve scheduling differently, or if the kids growing up with this stuff thought it was a good idea.
Photo by Zetong Li / Pexels
Gallup’s study on Gen Z is the canary: half use AI, but they’re getting angrier and less hopeful. That’s not a temporary sentiment. That’s the sound of a generation watching someone else open Pandora’s box and deciding they’re not going to thank them for it.
The irony? These same companies are suddenly pretending to care about guardrails. Anthropic says it’s “working with 40 companies” to explore cybersecurity applications for Mythos—their dangerous model. That’s not releasing it responsibly. That’s beta-testing harm with a fig leaf.
The SEO Hostage Situation
While AI labs play superintelligence poker, something weirder is happening at ground level: businesses are scrambling to get noticed by AI search.
This is the thing nobody talks about. Google’s AI Overviews pull answers from everywhere—trustworthy sites, Facebook posts, Reddit threads where someone made something up at 2 AM. Companies are now optimizing their entire web presence not for humans but for whatever algorithmic inkblot Google’s models produce. It’s SEO, but the search engine hallucinating is a feature, not a bug.
My read is this destroys information quality in ways we haven’t priced in yet. In 2024, you could still theoretically find authoritative sources if you knew what to look for. By 2027, everything’s been rewritten to please AI crawlers, and the original authority gets buried.
It’s like if every restaurant changed their menu in 2005 because Google Maps was becoming the primary way people found them. Except the menu now has to appeal to a machine that confidently confuses onion rings with calamari.
Photo by nappy / Pexels
The Privacy Incident That Should Scare You More
An ex-Meta worker downloaded 30,000 private Facebook photos.
Read that sentence again. Not a security breach. Not a hack. An employee who had access, with presumably legitimate reasons to access systems, just took 30,000 images and walked out the door. Meta fired them after discovering it. But notice what didn’t happen: you probably didn’t hear about it until now. It was a headline, not a congressional moment.
Five years ago, this would’ve been a three-day scandal. Now it’s Tuesday. Now it’s background radiation.
The real question is how many times this happens without getting caught. If one person took 30,000 photos, how many took 3,000? How many took nothing but looked? The meta-problem (pun intended) is that we’ve stopped being shocked by privacy incidents at scale. They’re just part of the operating cost now.
The Regulation Speedrun
Meanwhile, actual governments are actually doing something.
Greece is banning social media for under-15s starting next year. France and Spain already did similar moves. This isn’t a toothless announcement—it’s law. Enforcement will be messy. Kids will use VPNs. But the policy vector is clear: Europe’s decided the benefits don’t justify the harms, at least for the youngest users.
America’s still in the debate phase while Europe’s in the implementation phase. That’s a twenty-year pattern at this point, and it keeps biting us. We argue about privacy in 2025 while Europeans are already building the alternative infrastructure for 2026.
I think by 2027, the US either rushes to copy European regulation badly (creating a fragmented mess) or accepts that American kids are on platforms Europe banned, creating a two-speed internet. Neither outcome is good.
The Satoshi Mystery That Doesn’t Matter
A British computer scientist denied being identified by the New York Times as Satoshi Nakamoto, Bitcoin’s creator.
I mention this because it’s a perfect symbol of how much the culture has shifted. Bitcoin launched in 2009 as this revolutionary peer-to-peer currency system. By 2025, we’re still trying to figure out who invented it, and… nobody really cares. Bitcoin’s just a thing. It’s a store of value for people who think the Fed’s incompetent and a payment system for people who’ve given up on traditional banking. The mythology matters less than the utility.
It’s similar to how we’ve stopped asking “is AI good?” and started asking “which AI features do I need to implement to not get left behind?” The philosophical moment passed. We’re in the logistics phase now.
What I’m Actually Worried About
The four-day work week thing from OpenAI bothers me more than it should. Not because shorter weeks are bad—I love shorter weeks. But because it frames massive labor displacement as a benefits package. It’s “we built something that makes you less necessary, so here’s more free time to figure it out.” That’s not generosity. That’s creative bankruptcy.
Here’s what I think happens: by 2027, we’ll have a two-class labor market. Highly specialized people doing work AI can’t replicate, paid well. And everyone else either retrained (expensively) or cycling through gig work and four-day weeks at lower salaries. The OpenAI proposal isn’t a universal policy—it’s a perk for people who already work at AI companies.
The Gen Z anger makes sense. They’re watching their career options get compressed before they even graduate.
What I’m Watching
Anthropic’s 40 partner companies and the Mythos model release timeline. If Mythos stays locked for more than 18 months, it means the industry actually believes AI capabilities can be dangerous enough to withhold. If it gets released in Q2 2025, it means that was theater. That’s the test of whether “safety” language is real or marketing.
Google’s AI Overview accuracy data and the SEO optimization feedback loop. Watch for when Google releases accuracy metrics. If they don’t, that’s confirmation they know the answer is bad. And track whether news sites start filing complaints about algorithmic preference for simplicity over accuracy.
Gen Z engagement rates with AI-adjacent platforms through Q3 2025. Gallup showed sentiment souring. If usage also drops, we’re looking at actual rejection of this tech cohort, not just cynicism. That would be significant.
The first jurisdiction to actually enforce an under-15 social media ban. Greece moving first means we’ll see real-world friction: school problems, social isolation complaints, parental backlash, VPN adoption rates. That data will either validate the approach or kill it. Everything else is prediction until we see the actual enforcement.