Look, I’m three fingers deep into my morning bourbon, trying to make sense of OpenAI’s latest PR extravaganza. They just announced their new o3 model, and guess what? None of us peasants can actually use it. Classic.
You know what this reminds me of? That fancy whiskey bar downtown that keeps their top-shelf stuff behind bulletproof glass. You can see it, dream about it, but unless you’re part of their special “safety research” club, you’re stuck with rail liquor like the rest of us schmucks.
This announcement caps off their “12 Days of OpenAI” event - because apparently, one day of overhyped promises wasn’t enough. They had to stretch it out like a bartender watering down drinks during happy hour. We got treated to a parade of features: real-time vision (so AI can watch you stumble), ChatGPT Search (because regular search wasn’t confusing enough), and even a Santa voice for ChatGPT (because nothing says “authentic Christmas” like an AI pretending to be jolly).
But here’s where it gets interesting, folks. While OpenAI’s playing hard to get with their new toy, Meta’s over here slapping AI into Ray-Ban sunglasses. That’s right - now you can have an AI assistant judge your fashion choices in real-time. They’re even throwing in live translation features, so you can butcher multiple languages simultaneously while looking like a tech-bro tourist.
And speaking of things that shouldn’t be exposed to the public, UnitedHealth’s Optum division left their AI chatbot hanging out in the open like a drunk uncle at a wedding. Anyone with a web browser could access it. Sure, they claim no sensitive information was exposed, but it’s like leaving your bourbon cabinet unlocked and hoping your teenage kids won’t notice.
The real comedy here is watching these companies trip over themselves to announce bigger, better AI models. Google just dropped their “reasoning” model, probably scheduled their press release during their morning coffee when they heard about o3. It’s like watching a bunch of frat boys trying to one-up each other at the bar - “Oh yeah? Well, MY AI can think harder than YOUR AI!”
Meanwhile, Databricks is sitting on a fresh $10 billion funding round, deciding not to go public because “it’s dumb to IPO this year.” Finally, someone in this circus showing a hint of common sense. Though when you’re swimming in that much cash, I guess you can afford to be rational.
The kicker in all this? OpenAI’s now got a 1-800 number so you can call ChatGPT from a landline. Because apparently, what the world really needed was the ability to drunk dial an AI at 3 AM. “Hey ChatGPT, why did my ex leave me?” Just imagine the possibilities.
You want to know the truth about all these announcements? They’re like those New Year’s resolutions you make after too many shots - ambitious, probably unrealistic, and by February, everyone’s forgotten about them anyway.
But hey, what do I know? I’m just a guy who writes about tech while maintaining a steady blood-alcohol level. At least when my local bartender makes promises, they come with a free shot of whiskey.
Stay real, stay human, and remember - if an AI tells you it can reason better than you, it’s probably never had to figure out how to expense a three-martini lunch.
Yours truly from the back booth, Henry Chinaski
P.S. If anyone from OpenAI is reading this, I’m technically a safety researcher. I research the safety of various bourbon brands daily. Just saying.
Source: OpenAI announces new o3 model – but you can’t use it yet | TechCrunch