Everywhere you turn, GPT-4 pops up. The buzz is all over social media platforms, media outlets, and on everybody’s lips. At least, the tech-savvy ones, which I assume you are one of them.
You must have heard of ChatGPT, a chatbot that was launched on the 30th of November, 2022. ChatGPT crossed the 1 million users benchmark barely 5 days after its launch which was recorded as the first in the world.
Barely four months after now, we have a newer model, GPT-4 as an improvement for GPT-3.5 on which ChatGPT was initially built.
The major difference between the models is that because GPT-4 is multimodal, it can use both text and image inputs, whereas GPT-3.5 can only process text inputs.
In similarities though, both models are trained till September 2021. Therefore, events after September 2021 are foreign to these AI models.
What does the GPT even mean?
GPT means Generative Pre-trained Transformers.
What do they do? They are a type of deep learning model used to generate human-like text. Human-like because the goal is to make the computer understand and process texts just the way humans do.
Also, GPT can be used to:
- translate text to other languages
- answer questions
- summarize texts
- generate different content types that have to do with texts.
What GPT-4 is about
GPT-4 is the latest large language model created by OpenAI. It was launched on the 14th of March, 2023. It can accept both text and image input to produce text output.
Also Watch: Introducing GPT-4
Although users can still access the chatbot for free, only those who pay a monthly subscription for “ChatGPT Plus” will be able to try out the updated version of ChatGPT powered by GPT-4. This subscription goes for $20 per month.
This is a great improvement in the Artificial Intelligence field, especially in Natural Language Processing (NLP). NLP is a branch of Artificial Intelligence that is concerned with giving computers the ability to understand and process text and words in the same way human beings can.
The improvement in GPT-4 from GPT-3.5 touched three areas
- Visual Input
- Longer context
GPT-4 as the more creative model.
It takes note of your writing style and generates content accordingly. It can generate, edit, and iterate with users on creative tasks.
Its creativity can also be sensed in its responses to users’ questions. It can also be used to write codes.
GPT-4 as a visual input acceptor
The best advancement with GPT-4 is the visual input. GPT-4 can accept a prompt of text and images. Specifically, it generates text outputs (natural language, code, etc.) given inputs consisting of interspersed text and images. Over a range of domains including documents with text and photographs, diagrams, or screenshots.
You could input a picture of ingredients and ask it to identify them and develop recipes from them, and it would do just that. Sounds cool, right? That is how advanced our technology is becoming.
GPT-4 as a capable of handler of longer context.
It can handle 25,000 words of text. You can use it for document analysis, script writing, and every other long writing. As opposed to the previous model which couldn’t take up to that.
According to OpenAI, GPT-3.5 has a maximum text length of 3000 words. To put it differently, the model is only able to reference up to 3000 words in a current conversation, before you will have to create a new chat.
However, this text limit can be split up into multiple messages and it is not the maximum length of a single message you can send Chat GPT. Also, the longer the text you input, the longer its processing time.
What people are saying about GPT-4
People’s reviews from across all popular social media platforms show that they are enjoying this new development. Users say the technology is incredible and transformative.
A Twitter User “Rowan Cheung” asks the AI to tabulate 20 human jobs it would replace and below is its response.
The question of whether it really would completely replace human jobs or not is a topic for another day.
Another Twitter User “Sudharshan” gave GPT-4 a photo of his stocked-up refrigerator for the AI to ientify everything inside it. Below is the input its response.
After seeing these examples, you will agree with me that the rave about GPT-4 is worth it.