Enter your keyword

ChatGPT and AI (2nd part of a series of posts about ChatGPT in particular and AI in general)

ChatGPT and AI (2nd part of a series of posts about ChatGPT in particular and AI in general)

https://openai.com/blog/chatgpt/

The popularity of ChatGPT, which has quickly reached millions of users, has worried Google's top brass. ChatGPT has delivered surprise after surprise, from its ability to answer questions in human-friendly natural language, correcting its own answers if deemed incorrect, composing poetry, helping students write essays based on their lecturers' assignments, solving simple math and engineering problems, co-authoring journal papers, passing 'bachelor' exams in several fields, and perhaps more surprises will follow. Although it still has several weaknesses, as a language model developed from GPT3.5 by combining supervised learning & reinforcement learning, ChatGPT has the potential to be retrained with datasets in more specialized fields, such as medicine, law, finance, etc., to make it smarter in these more specialized fields. Currently, the ChatGPT model is being trained with datasets sourced from the Internet using WebCrawler and books (some e-books?) until 2021.  GPT stands for Generative Pretrained Transformer. This acronym implies that ChatGPT is a language model pre-trained on a language dataset, enabling it to acquire intelligence based on the Transformer model. Transformer is a deep learning model currently very popular in natural language processing (NLP) and has recently also been applied to computer vision. 

CEO Google Sundar Pichai  They even had to call Google founders Larry Page and Sergei Brin to discuss strategies to respond to the popularity and threat of ChatGPT. It is estimated that in the not too distant future, the Google Search Engine will be equipped with an AI Chatbot like ChatGPT, so it seems that not only can users dialogue with the Chatbot, but the Chatbot itself can search for necessary and relevant information on the Internet. As you may already know, Google is also developing a dialogue language model called Lambda & Sparrow. Recently, Google even claimed to have developed an AI model that generates photos, videos, and audio controlled by natural conversational language, photos, and videos. It will be interesting to see whether Google and OpenAI will compete fiercely in the future in the field of AI, whether in the fields of natural language processing, chatbots, or computer vision. 

It goes without saying that all these rapid developments and advancements in AI will bring about transformations in how we work, learn, communicate, and socialize. How do we, as educational institutions, respond to various developments in the field of AI, such as ChatGPT? Mr. Kuspriyanto once asked ChatGPT about this, and ChatGPT provided several suggestions/recommendations. Do we need to discuss and further explore ChatGPT's suggestions/recommendations? In this context, I am reminded of the message conveyed by Nick Polson & James Scott in their book entitled "Artificial Intelligence Quotient: How People and Machines Are Smarter Together" (I bought this book at the MIT Cooperative Store during a recent visit). They wrote, "A revolution of intelligent machines, from self-driving cars to smart digital assistants, is now remaking our world, just as the Industrial Revolution remade the world of the nineteenth century. AI is not some science fiction droid from the future. It's right here, right now, and it's changing our lives at lightning-fast speed." 

Bandung, 30 Januari, 2023

Bambang Riyanto