Why do people criticize GPT-4 vision so much? It’s amazing, but I keep seeing people posting things like, “Look, I can show you where it isn’t perfect, so it isn’t actually cool or useful”. I don’t think we realize just how far we’ve come so quickly.
GPT-2 was amazing. There were multiple paragraphs written that (usually) you couldn’t tell was written by AI. It was amazing.
GPT-3 blew that away. It reliably output human-sounding text, and you could start having back-and-forth conversations
GPT-4 - I had no expectation of this being anything more than a marginal improvement. I was so wrong.
GPT-4 Vision will change the world. Machines can now process the world like we do - visually and with audio/text. Nothing is impossible. We've been playing around with this, and trust me, no other multi-modal model is even close. We're talking orders of magnitude different in terms of real use cases.
For me the real kicker is cost - so many cool tools that can transform so much of what we do, but how can you use them without killing your business?
I’m excited for OpenAI’s Nov. 6 dev days and their announcement about lowering costs and 🤞 hopefully the API for GPT-4Vision.
