Artificial intelligence has seen significant growth in recent years, driven by advancements in Machine Learning and Natural Language Processing (NLP). One of the most notable developments in NLP is the rise of large language models such as ChatGPT.
ChatGPT, which stands for "chat-based Generative Pre-training Transformer" is a language model developed by OpenAI that is trained to generate human-like text. Over the last three years, ChatGPT was trained on 175 billion parameters and 300 billion tokens, equivalent to 570 GB of content from sources like Wikipedia, social media, news, and articles. The real breakthrough of ChatGPT is its user-friendly, open-source model that is freely available, making it a widely accessible, general-purpose sophisticated chatbot capable of writing like a human.
OpenAI has partnered with several companies to integrate ChatGPT into their applications. Microsoft, who invested $1bn in OpenAI three years ago, recently announced an additional $10 billion investment to incorporate ChatGPT into its products and make it widely accessible. Microsoft can use ChatGPT in a variety of ways, some examples including NLP for language understanding in products such as Cortana, Microsoft Teams, and Office; Text generation for products such as Word, PowerPoint, and Outlook to assist with writing and formatting documents and emails; Language Translation for products such as Microsoft Translator and Microsoft Edge to improve the accuracy and fluency of its language translation and make communication between users with different languages more seamless; Chatbot for customer service to develop a more natural and efficient communication between users and the products; and more.
With the increase in popularity of ChatGPT, many have started wondering if this could threat or replace Google’s search engine business. ChatGPT and Google have some similarities, such as a user-friendly interface and both help users get valuable information. However, it’s important to note that ChatGPT is not designed to search and retrieve information like Google's search engine does: The model is more focused on generating text based on a given context, but it does not have the capability to find and index the information from the internet. Also, while ChatGPT is focused on generating answers based on specific input, Google offers a wider range of services and products like cloud storage, email, and more. Although ChatGPT can assist users with tasks, it is unlikely to pose a significant challenge to Google in the near term as they serve different purposes and have different capabilities.
In addition, Google has also developed proprietary LLMs that are as advanced or more, than GPT, but have not made it available on an open-source basis. LaMDA, which stands for “Language Model for Dialog Applications” is a supervised-learning model and a “Google-equivalent” of ChatGPT. LaMDA is more accurate and broader in scope than ChatGPT, as it is trained on 540billion parameters, 3x that of ChatGPT. Also, Google uses a vast amount of internal data, giving LaMDA access to a larger data set to train its model compared to competitors, which gives Google a significant advantage in terms of quality and diversity of data available for training.
The future for ChatGPT is clearly promising: The model took only 5 days to hit its first 1 million users, while Instagram and Netflix took 3 months and 3.5 years respectively to reach the same target, making it the fastest-growing tech platform ever. As of today, the model is estimated to have an error rate of 2-5%, according to Bloomberg. But as with any technology, the costs and errors associated with using ChatGPT are likely to decrease over time as the model is further developed and optimized. The growth and popularity of ChatGPT are not expected to end soon, as the technology continues to advance, and more companies and organizations look to integrate the model into their products and services.
The information in this article should not be regarded as a description of services provided by Delian Partners SA. The opinions expressed in this article are for general informational purposes only and are not intended to provide specific advice or recommendations for any individual or on any specific security or investment product. It is only intended to provide education about the financial industry. The views reflected in this article are subject to change at any time without notice.