The Grim Reason OpenAI Withheld the New ChatGPT Version: Bespoke Pornography
They promised voice and video for GPT-4o but never delivered
“It’s OpenAI’s version of Her!”
“This changes EVERYTHING!”
“And it comes with Scarlett’s voice!!!”
That’s what influencers, journalists, and analysts shouted back in May when OpenAI did the GPT-4o demo. I was also excited about it: an uncannily human voice, responsive to emotions, real-time audio/video interaction, and better at math and code than GPT-4 Turbo. The demo was sick.
Now we know it was criminally rushed.
In a final attempt to win the battle against Google—the demo aired the day before Google’s I/O conference—OpenAI made its biggest mistake yet, they promised more than ever and delivered nothing.
This is the question everyone’s been asking for two months now: Why aren’t they rolling out the voice and video features? What’s holding them back?
There’s an official reason—the one OpenAI gave journalists from the NYT and Bloomberg—and an unofficial one. If you’re like me, you know you can skip the NYT piece.
The other reason, well, it didn’t quite surprise me. But still, I’d rather not know what’s coming.