GPT-4 An Advancement For AI, Question Is How Much

It all started on a chilly San Francisco December day in 2015, a company that took it upon itself to revolutionalise the landscape of Artificial Intelligence (AI) in the world while maintaining ethical boundaries was born, unknown to (almost) none today - OpenAI.

- Oishani Bhattacharya

Five heads, Sam Altman, Greg Brockman, Ilya Sutskever, Wojciech Zaremba and Elon Musk, got together to build OpenAI with the belief that AI has the potential to help solve some of the world's most pressing problems.

In November 2022, the said firm introduced humankind to ChatGPT, an AI-powered chatbot built on Generative Pre-Trained Transformer Architecture (GPT). The AI chatbot has been trained on a huge amount of textual data from the internet - a type of neural network which was introduced in 2017 in a research paper titled, "Attention Is All You Need." The chatbot takes prompts in the form of texts and gives the user a detailed answer based on that essays, poems (rather ghastly ones but yeah), computer codes, songs, screenplay, etc.

Cut to March 2023, and now we have GPT-4, an even bigger and more intelligent model (aced every single test humans have ever taken and passed with flying colours) than ChatGPT. Its USP is that, unlike ChatGPT which only took prompts in the form of text, GPT-4 accepts image inputs as well. Since its launch, many users have subjected GPT-4 to several images (random ones to even memes) and the AI model was seen accurately (in most cases) interpreting the image and even explaining why a certain image is considered funny.