The New GPT-4: What's Next for Language AI?
Introduction
OpenAI released GPT-4 on March 14th, 2023. It is OpenAI’s most advanced system, producing safer and more useful responses.
It’s already available in ChatGPT if you’re a pro subscriber, and API access is behind the waitlist.
It is already being used by clients ling Stripe, Duolingo, Descript, and much more
Major Upgrades
Here are the major upgrades in GPT-4
1. Visual?Input
GPT-4 is a Multi-Modal Model which means that it can accept images as inputs and generate captions, classifications, and analyses.
For example, you can take a picture of the ingredients that you have and it will tell you what you can cook with them.
2. GPT-4 is?Smarter
It passed the bard exam in the top 10% unlike GPT-3 which was in the bottom 10%.
It can now solve most of the easy questions on Leetcode, but it still fails for the medium and hard questions.
3. Better Understanding of?Context:
GPT-4 can handle over 25,000 words of text compared to about 3,000 for GPT-3.5, allowing for use cases like long-form content creation, extended conversations, and document search and analysis.
领英推荐
This means it will be able to understand the meaning behind the words and generate more relevant responses to the input given.
You can provide the whole documentation of any library and it will be able to answer all of your questions about it. It can do the opposite job as well where you pass a code and it writes documentation for that code, you can also use it to analyze your code or translate code from one language to another and soon it will be integrated into Copilot.
4. More Parameters:
One of the most obvious upgrades in GPT-4 is an increase in the number of parameters.
GPT-3 already has 175 billion parameters, GPT-3.5 has 190 billion parameters and GPT-4 has even more. GPT-4 parameter details are undisclosed but rumored to be around 100 trillion.
And with more parameters comes improved accuracy. GPT-4 has a better understanding of language and is able to generate more human-like text. This means fewer mistakes and more coherent responses.
5. Multilingual Capabilities:
Another upgrade in GPT-4 is improved multilingual capabilities. GPT-3 already supports multiple languages, but GPT-4 can take this a step further and be able to understand and generate text in even more languages.
6. More Creativity
GPT-4 is more creative and collaborative. It can generate, edit, and iterate with users on creative and technical writing tasks, such as composing songs, writing screenplays, or learning a user’s writing style.
Conclusion
In conclusion, GPT-4 is a significant upgrade over GPT-3.5. With more parameters, improved accuracy, multilingual capabilities, and a better understanding of context, we can expect GPT-4 to be one of the most advanced language models yet.?
However, the drawbacks are that it’s noticeably slower in generating responses and still has many known limitations such as social biases, hallucinations, and adversarial prompts.