ChatGPT, Ένα σύγχρονο πρόγραμμα επεξεργασίας και κατανόησης φυσικής γλώσσας : τεχνολογία και εφαρμογές
ChatGPT, A modern program for natural language processing and understanding : technology and applications
View/
Keywords
ChatGPT ; Large Language Models (LLMs) ; Transformer architecture ; Natural Language Processing (NLP) ; Deep learning ; Reinforcement Learning with Human Feedback (RLHF) ; Tokenization ; Artificial Intelligence (AI) ; Ethical implications ; Human-machine communicationAbstract
This master's thesis explores ChatGPT as a cutting-edge system for natural language understanding and processing, focusing on the underlying technologies and its real-world applications. Built upon large language models (LLMs) and the Transformer architecture, ChatGPT represents one of the most advanced examples of artificial intelligence in the field of natural language.
The study begins by presenting the theoretical foundations of natural language and artificial intelligence, emphasizing the evolution of Natural Language Processing (NLP). It analyzes the neural network and deep learning techniques used in the development of LLMs. A detailed examination follows, focusing on the architecture, operation, and training techniques of ChatGPT, including supervised learning, unsupervised learning, and reinforcement learning from human feedback (RLHF). Key concepts such as tokenization, subword encoding, fine-tuning, and activation functions are also discussed.
The thesis then investigates the practical applications of ChatGPT across various domains, including education, business, journalism, scientific research, and the legal and medical sectors. Special attention is given to the ethical implications and challenges surrounding its use, such as misinformation, bias, and data privacy.
Finally, the work discusses future directions of this technology, potential improvements to language models, and the evolving role of AI in human-machine communication.