Artificial Intelligence (AI) is a multidisciplinary field concerned with the development and deployment of intelligent machines capable of emulating and executing tasks that typically require human cognitive abilities. The concept of intelligent machines predates modern computers by thousands of years, with ancient myths and folklore across various civilizations depicting tales of artificially created beings with human-like intelligence, such as the bronze automaton (self-operating machine) Talos from ancient Greek mythology. AI has also been a popular topic in science fiction since the early 20th century.
The study of AI as a scientific discipline began to take shape in the 1950s. In 1956, a workshop held in Darthmouth College (USA) marked a significant milestone, as a group of researchers coined the term "artificial intelligence" and laid out the groundwork for AI research and development. Early AI pioneers focused on developing programs capable of problem-solving and logical reasoning. However, progress was relatively slow due to limitations in computational power and the complexity of representing and manipulating knowledge. By the 1970s, it became apparent that researchers had greatly underestimated the difficulty of the project. This led to a period of decline and decreased funding in artificial intelligence research, also known as the "AI winter".
The field of AI experienced a significant resurgence in the late 1980s, driven by advancements in machine learning and the availability of more powerful computers. Machine learning, a subfield of AI, encompasses algorithms and techniques that allow machines to learn patterns and make predictions or decisions based on data. In particular, the development of artificial neural networks and their ability to learn from examples revolutionized the field. The emergence of expert systems, which leveraged knowledge-based rules and inference engines, also contributed to the progress of AI applications in various domains.
Neural networks, computational models inspired by the human brain, have witnessed remarkable advancements over the years and come in various types. Convolutional Neural Networks (CNNs), introduced in the 1980s, revolutionized image and video processing by extracting spatial features. Recurrent Neural Networks (RNNs), developed in the 1990s, excel in handling sequential data with their feedback connections, making them ideal for tasks such as language modeling and sentiment analysis. Deep learning emerged as a powerful approach within the field of artificial intelligence, and it gained significant attention in the mid-2000s. Deep learning involves training neural networks with multiple layers (hence the term "deep"), allowing them to automatically learn and extract hierarchical features from data. This approach enabled neural networks to effectively model complex patterns and relationships in various types of data. Generative Adversarial Networks (GANs) and Variational Autoencoders (VAEs), emerging in the mid-2010s, pushed the boundaries of generative modeling, allowing the generation of realistic data samples. Transformers, introduced in 2017, have gained prominence in natural language processing (NLP) by capturing long-range dependencies within sequences effectively. The transformer architecture has since become the backbone of numerous state-of-the-art models in NLP, including the Generative Pre-trained Transformer (GPT) language model series.
The 2020s have witnessed a significant breakthrough in generative AI, marked by the introduction of transformative tools such as ChatGPT and other large language models (LLMs), image generation tools like DALL-E 2, Midjourney, and Stable Diffusion and other generative AI tools (speech, music, videos, 3D models). ChatGPT, launched in November 2022, is a new type of chatbot built on the highly advanced GPT-3.5 and GPT-4 large language models. This launched a huge public interest towards the technology, with an estimated number of 100 million monthly active users just two months after its launch. ChatGPT is based on the Transformer architecture and has been trained on a diverse range of internet text to generate human-like responses in natural language conversations. ChatGPT is designed to understand and generate text based on the provided context, making it well-suited for chatbot applications. It can assist users in answering questions, providing information, engaging in conversation, and offering suggestions. Alongside the launch of ChatGPT, the OpenAI API was made available to all developers worldwide, leading to a surge in new projects that utilize the technology. Additionally, many open source LLMs, such as Meta's LLama 2, are freely available. This has resulted in thousands of new LLM-based services, tools, frameworks and projects in a relatively short period of time.
Artificial Intelligence (AI): A multidisciplinary field focused on developing and deploying intelligent machines capable of performing tasks that typically require human cognitive abilities.
Chatbot: An AI program or system designed to interact and communicate with humans through text or speech.
Deep learning: A subset of machine learning that involves training artificial neural networks with multiple layers to analyze and learn patterns from data. Deep learning has been instrumental in achieving breakthroughs in various AI tasks, such as image and speech recognition.
Generative AI: A form of artificial intelligence that creates content (e.g. text, images, audio or video) based on its training data.
Generative Pre-trained Transformer (GPT): A language model based on the transformer architecture, developed by OpenAI, known for its ability to generate coherent and contextually appropriate text.
Large language models (LLMs): Advanced language models that have been trained on vast amounts of text data and are capable of generating human-like responses and engaging in natural language conversations.
Machine learning: A subfield of AI that focuses on developing algorithms and techniques enabling machines to learn from data and make predictions or decisions without being explicitly programmed.
Natural Language Processing (NLP): A subfield of AI concerned with enabling computers to understand, interpret, and generate human language.
Neural networks: Computational models inspired by the structure and functioning of the human brain, used in machine learning and AI to process and analyze data.
Prompt: A question or instruction given to the AI to produce a desired output.
Prompt Engineering: Optimizing your prompt using the right words, phrases, formatting, context etc. to get better quality and relevant outputs from the AI.
Transformer: A type of neural network architecture introduced in 2017 that has proven highly effective in capturing long-range dependencies within sequences, particularly in natural language processing (NLP) tasks. Popular text generators such as ChatGPT are based on the transformer architecture.
OpenAI API: An application programming interface provided by OpenAI that allows developers to access and utilize the capabilities of OpenAI's language models and other AI technologies.