The other day I saw this video - its from 4 ish months ago - Eric Schmidt - ex Google and his insights are addictive and accurate.
Got me thinking that the door is officially open. But also important to understand below surface value.
The $5.5M AI darling that supposedly leapfrogged OpenAI and LLaMA. Spoiler: That $5.5M? Total media spin. The real cost? You still needed OpenAI and LLaMA to get built. What did they do? Scrape a model, optimize hardware, and call it “innovation.”
We all knew someone would try this. They just got there first.
How They Did It:
DeepSeek didn’t invent something new; they duct-taped together existing frameworks, made it run cheaper, and called it a day. Honestly? It’s kind of impressive. But the quirks in their outputs? Yeah, those scream “borrowed.”
Here’s where it gets Spicy:
GPU export restrictions? They found a workaround. Data privacy? Their terms reportedly send data straight to China. But let’s be fair, the U.S. is no saint either. We just make it feel better by keeping our data theft “Made in the USA.”
What’s Actually Interesting:
Forget the drama, the real breakthrough is making AI cheaper and more accessible. That’s a big deal. But this also raises the stakes: AI is now powerful enough to turn “build me a TikTok clone” into a one-hour project. The possibilities are exciting...and a little unsettling.
My POV:
DeepSeek is a reminder that innovation often starts with repurposing and optimizing what’s already there. But it also highlights the need for accountability. Without boundaries, we risk losing trust in the tech we’re building. Actually we already are...the problem is that people dont care until it affects them directly and immediately (see the next 4 years).