The new AI revolution arrived sooner than we expected with GPT-4! Bard or not, OpenAi is not playing around and is heading for the crown.
When ChatGPT-3 was introduced, everyone thought that it was the beginning of the end for humans. With its coding, writing, translating, organizing skills, etc., etc.… it has certainly helped us, but also made us a bit lazy as well. Now, we have the upgraded version of it. So, have we reached the finish line already?
A glimpse of GPT-4
First, let’s get to know GPT-4 first. Is it really that different from its predecessor? Let’s see!
It was introduced on March 14, 2023, and it’s the new hit thing in deep learning. Yes, deep learning just got deeper! While GPT-3 can summarize texts, chat, give human-like answers, and is a great help in many ways, it’s not always accurate.
GPT-4, on the other hand, provide more accurate responses with its advanced reasoning capabilities. Unlike GPT-3, this one can understand beyond words. Yup, it can process both words and images in addition to everything else its predecessor can do. So obviously it was trained on a much larger dataset; 45 GB to be exact. That’s 28 GB more than GPT-3. They both share one thing in common though, and it’s that both have limited knowledge of events after 2021.
Data says...
OpenAI revealed that GPT-4 passed “a simulated bar exam with a score around the top 10% of test takers; in contrast, GPT-3.5’s score was around the bottom 10%.” It also outperformed existing large language models, alongside most state-of-the-art (SOTA) models.
The new model can change images into text, tell you why an image is funny, find loopholes in contracts
So far, the new model GPT-4 is only available for ChatGPT Plus members, so it’s not as accessible and thus, cannot tell how much better is it than GPT-3.
