When ChatGPT was released last year on November 30 by Open AI. It made waves across the internet. And on 14th march 2023 they have announced its new version GPT-4.Whichis more advanced than ChatGPT.

But before we proceed, first let’s understand how both of them work:

ChatGPT is an AI language model that uses a neural network called a transformer to predict the next word in a sentence given the preceding words. It is trained on a large corpus of text data and can be used for a variety of natural language processing tasks. When a user inputs text, ChatGPT processes it and generates a response based on the patterns it has learned from the text it was trained on. The quality of the responses depends on the input and context, and while it can produce high-quality responses, it is not perfect and may sometimes produce nonsensical or inappropriate responses.

Unlike its predecessor GPT-4 can take images as well as text as input; this gives it the ability to describe the humor in unusual images, summarize screenshotted text, and answer exam questions that contain diagrams.

Now What does GPT-4 stand for?

GPT-4 stands for Generative Pre-Trained Transformer 4

GPTs are machine learning algorithms that respond to input with human-like text. They have the following characteristics:

• Generative: They generate new information.

• Pre-trained: They first go through an unsupervised pre-training period using a large corpus of data. Then they go through a supervised fine-tuning period to guide the model. Models can be fine-tuned to specific tasks.

• Transformers: They use a deep learning model -- transformers -- that learns context by tracking relationships in sequential data. Specifically, GPTs track words or tokens in a sentence and predict the next word or token.

While ChatGPT offers a number of advantages, there are also some potential draw backs that should be considered when using it. Here are some of the key drawbacks of ChatGPT:

• Lack of context: ChatGPT may generate responses that are accurate but lack the appropriate context, leading to confusion or misunderstandings.

• Limited creativity: While ChatGPT is able to generate high-quality text, it may struggle to generate truly creative or innovative responses that go beyond the patterns and structures present in the training data.

• Limited knowledge: While ChatGPT has access to a vast amount of information, it is still limited by the quality and quantity of the data on which it was trained, and may struggle to provide accurate or helpful responses to highly specific or technical queries.

With GPT-4 all the drawbacks have been resolved to a greater extent, thanks to its broader general knowledge and problem solving abilities.

‍

So let’s look at the areas where GPT-4 improved :

• Creativity: GPT-4 is more creative and collaborative than ever before. 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.

‍

• Visual Input: You can now give an image as an input, and GPT-4 can generate captions, classifications and even analyze them.

‍

• Longer Context: GPT-4 is capable of handling over 25,000 words of text, allowing for use cases like long form content creation, extended conversations, and document search and analysis.

Now that we’ve seen its renewed capabilities let us look at some of the previous task people have asked ChatGPT to perform certain tasks or exams and see for ourselves the brevity of Open AI’s highlighted statement, “GPT-4 surpasses ChatGPT in its advanced reasoning capabilities”.

Here’s a demonstration of ChatGPT’s response about a situation asking when three people can meet.

‍

Then the same prompt was given to GPT-4 and here’s the response:

‍

By looking at both the responses, we can see that GPT-4 is less wordy and to the point.

‍

ChatGPT vs GPT-4 exam comparisons in terms of percentile.

‍

Gaining popularity: With that kind of improvement, GPT-4 has already garnered attention of many companies and some of them have already collaborated and launched their products such as Khan Academy’s AI powered chatbot which acts as an online tutor.

And Duolingo’s AI powered Duo Max which has two new features: Role Play, an AI conversation partner, and Explain my Answer, which breaks down the rules when you make a mistake, in a new subscription tier called Duolingo Max.

Apart from these two popular companies many others like stripe, Morgan Stanley and the Icelandic govt. Etc. are incorporating GPT-4 to strengthen their capabilities.

Following the research path from GPT, GPT-2, and GPT-3, Open AI’s deep learning approach leverages more data and more computation to create increasingly sophisticated and capable language models.

However they still believe there is more to improve as the company stated: “GPT-4 still has many known limitations that we are working to address, such as social biases, hallucinations, and adversarial prompts. We encourage and facilitate transparency, user education, and wider AI literacy as society adopts these models. We also aim to expand the avenues of input people have in shaping our models”.

We at Alphaa AI are on a mission to tell #1billion #datastories with their unique perspective. We are the community that is creating Citizen Data Scientists, who bring in data first approach to their work, core specialisation, and the organisation.With Saurabh Moody and Preksha Kaparwan you can start your journey as a citizen data scientist.