I have obsessed over a lot of new tech over the years with one of them being ChatGPT. I have been all over it since version 2 was released many years ago. I even mentioned how it’s going to replace developers and administrative roles in a past post.
Now that GPT-3 has been released, I’m underwhelmed. I haven’t discussed this with anyone else yet but I was so disappointed that this is the best version release after so many years. It’s slow. The results while more comprehensive aren’t all that different from GPT-2. The last version, I wasn’t able to source any of the results provided and I really hoped this version I’d be able to, but no dice. There still aren’t any tracing or image features. If I entered, “Show me the trading fibonacci retracement of bitcoin in 2022” and it actually gave me nearly identical text definition as GPT-2. I was hoping for a visual but I guess this might not be the version to do it.
So what does GPT-3 do really well is provide much better results and the write ups are incredibly coherent now. I feel like myself and most other people could be fooled in thinking a person wrote a poem, an article, or school essay written by AI. Actually, not a school essay. Version two struggled with this and it failed the plagiarizer. I just tried running a GPT-3 generated essay into Google’s plagiarizer and it identified over 90% plagiarized. So… yeah…
Basically there’s still a long way to go. I am disappointed in the progress of OpenAI because so many years have gone by. I was really hoping for significant change. At this rate, OpenAI won’t be replacing jobs, search engines, or analysis work anytime soon. Skynet will have to wait.
