Strong disagree. My personal number has also leaked and I get 5-10 cold calls per day. They call during important work meetings. They call during dinner with my family. They call when I'm away on vacation. I will never, ever willingly buy any product from these calls.
Don't know when that will get here, but when it does (and better AI-powered spam filtering), I think a lot of these problems will disappear, and a lot of sales teams are going to have to rethink.
I’d bet it’s using function calling out to a real chess engine. It could probably be proven with a timing analysis to see how inference time changes/doesn’t with number of tokens or game complexity.
?? why would openai even want to secretly embed chess function calling into an incredibly old model? if they wanted to trick people into thinking their models are super good at chess why wouldn't they just do that to gpt-4o?
The idea is that they embedded this when it was a new model, as part of the hype before GPT-4. The fake-it-till-you-make-it hope was that GPT-4 would be so good it could actually play chess. Then it turned out GPT-4 sucked at chess as well, and OpenAI quietly dropped any mention of chess. But it would be too suspicious to remove a well-documented feature from the old model, so it's left there and can be chalked up as a random event.
Cool, I built a prototype of something very similar (face+voice cloning, no video analysis) using openly available models/APIs: https://bslsk0.appspot.com/
The video latency is definitely the biggest hurdle. With dedicated a100s I can get it down <2s, but it's pricy.
With a dedicated GPU and some cleverness it can be relatively quick. I split the response on punctuation and generate smaller clips in a pipeline. I haven't taken the model apart to try streaming the frames coming out of ffmpeg yet, but that would probably help a lot.
This was my undergraduate "thesis" project. We built a GDB plugin that would generate a puredata audio structure to help with debugging. Eg. it would play snoring during a sleep() call, each breakpoint could be set to a different frequency, etc. It was actually pretty interesting for multi-threaded code, though pausing to play the audio could change the runtime profile.
I have a lot of examples but a funny one that comes to mind is: in the early 2000s when IM clients were all the rage, I wrote a VB6 application to go through my MSN Messenger logs and rank my friends by how much I talk to them. Kind of like a MySpace top 10 prior to MySpace.
I spent a decent amount of time tweaking the UI, improving performance, adding filters, providing different file output formats, etc. Never shared it with anyone.
Also, at least the last time I checked, the Google Maps API was exorbitantly expensive. Niantic would likely get a discount, but avoiding this exploit doesn’t override all of the other benefits of OSM.
Similarly the Canadian province of Ontario leased it's only toll highway for 99 years to an international group, for an amount it currently earns in tolls every ~2 years. Fun.
You're missing the bigger picture though. By doing so they managed to look like they had balanced the budget in an election year. That's what really matters here.
Fair. At the current rate, the revenue takes two years to cover the original cost. The net profit takes 6. The margins are quite good due to consistent price increases.
