🥎 Important Terms

Abstract

This chapter covers some of the important terms related to Generative AI and Prompt Engineering.

🟥 Artificial Intelligence

Artificial Intelligence is a field of computer science which deals with building machine which can learn without being explicitly programmed. These machines use algorithms and data to learn and do tasks like reasoning, problem-solving, and decision-making which require human intelligence. Artificial intelligence includes several areas like natural language processing, computer vision, speech processing, machine learning and robotics.

🟧 Machine Learning

Machine Learning is an area of artificial intelligence which focuses on developing algorithms that enable machines to learn by identifying the patterns in the data. Machine learning avoids the machines being explicitly programmed and offers the learning ability. Machine learning involves areas like deep learning, reinforcement learning etc.

🟨 Natural Language Processing

Natural language processing (NLP) is an area of artificial intelligence which focuses on developing algorithms that enable machines to understand, interpret and generate natural language. NLP enables machines to do many tasks like text classification, entity extraction, question answering etc. which fall under natural language understanding and tasks like text summarization, machine translation, text generation etc. which fall under natural language generation.

🟩 Computer Vision

Computer Vision (CV), an area of artificial intelligence develops algorithms that allow machines to understand, interpret and generate visual information (images, videos). CV enables machines to do many tasks like image classification, video classification, object detection etc. AI-based image and video editing applications are developed using deep learning and computer vision.

🟦 Speech Processing

Speech Processing is an artificial intelligence area focusing on understanding and generating speech. It involves various tasks like speech recognition (converting spoken words into text), speech synthesis (generating speech from text), and speech signal processing (analyzing and enhancing audio signals containing speech). Popular virtual assistants like Apple’s Siri, Amazon’s Alexa etc use speech processing techniques to understand the user commands.

🟪 Deep Learning

Deep Learning is an area of machine learning which focuses on developing algorithms which doesn’t require explicit feature engineering and learn features hierarchically using deep layered networks. Deep learning avoids the time and labor-intensive process of feature engineering which also requires domain expertise in many cases. Popular chatbots like ChatGPT and Bard are built based on advanced deep learning models like Transformers.

🟦 Generative Artificial Intelligence

Generative Artificial Intelligence is an area of artificial intelligence which focuses on developing algorithms that can learn from existing data and then generate new data. Generative AI models can generate text, image, video or speech data. Generative AI is at the intersection of other AI areas like natural language processing, computer vision, machine learning and speech processing. Popular AI applications like ChatGPT and Bard are examples of Generative AI.

🟫 Transformer

Transformer is an advanced deep learning built on top of self-attention mechanism. Transformer which completely avoids recurrence and convolution layers is more parallelizable and hence can better leverage advanced computer hardware like GPUs and TPUs. The transformer is the backbone behind pretrained language models like BERT, RoBERTa, T5 etc and large language models like GPT-3, GPT-4, Gemini etc.

🟪 Pretrained Language Model

A pretrained language model is an advanced deep learning based on transformers and pretrained on large volumes of text data. Some of the popular pretrained language models are BERT, RoBERTa, ELECTRA, DeBERTa etc. To summarize, think of the pretrained language model as a well-trained model which can be fine-tuned for downstream tasks with comparatively less labelled data.

🟦 Large Language Model

Large language models (LLMs) are a special class of pretrained language models obtained by scaling model size, pretraining corpus and computation. LLMs, because of their large size and pretraining on large volumes of text data, exhibit special abilities which allow them to achieve remarkable performances without any task-specific training in many of the natural language processing tasks. Some of the popular examples of LLMs are GPT-3, GPT-4, Llama2, Gemini etc. To summarize, think of the large language model as a well-trained advanced model which can used directly without fine-tuning for downstream tasks.

🟥 Prompt

A prompt is a request given to the chatbot to get the desired output. Here the request is given in natural language i.e. as text input. Here the text input can be a query asking for some information or an instruction asking the chatbot to do a task. To summarize, a prompt is request in natural language to the chatbot.

🟧 Prompt Engineering

Prompt Engineering refers to the process of crafting and refining prompts iteratively to get the desired output from the chatbot. Prompt engineering helps to improve the prompts in successive attempts to get an output which aligns with the user requirements. To summarize, prompt engineering helps to refine the prompts and effectively leverage the abilities of the chatbot to get the expected output.

🟩 GPT-3

GPT-3 is the first LLM which attracted both academic and industry people with its exceptional abilities to do tasks without any explicit task-specific fine-tuning. The GPT-3 model was introduced by OpenAI in May 2020 and later it was further improved. GPT-3 model consists of 175 billion parameters and trained over huge volumes of text data from a variety of sources like Books, Webpages, Wikipedia etc.

🟦 GPT-4

GPT-4 is the latest and most advanced LLM introduced by OpenAI. GPT-4 is much bigger and trained on a much larger text corpus compared to the GPT-3 model. The size and other details of the GPT-4 model are not revealed by OpenAI. To summarize, GPT-4 is the latest and largest LLM introduced by OpenAI capable of handling both text and image inputs.

🟪 Chatbot

A Chatbot is a computer program developed to simulate human conversation, using text or voice interactions, to answer questions, or assist with tasks. It uses deep learning algorithms and natural language processing techniques to understand and generate responses conversationally. Chatbots are commonly used in customer support, virtual assistants, and various other applications to automate and enhance communication with users. Some of the popular examples of chatbots are ChatGPT, Bard etc.

🟫 ChatGPT

ChatGPT is an advanced chatbot powered by GPT-3.5 and GPT-4 models. ChatGPT was launched for public use in November 2023. ChatGPT can be utilized in a wide range of applications, including customer support, content generation, as a virtual assistant etc.

🟥 Gemini

Google's Gemini is a cutting-edge generative AI model designed to understand and generate different types of information like text, code and images. Gemini was introduced by Google in December 2023. Gemini Pro can handle text inputs while Gemini Pro Vision is multimodal and can handle both text and image inputs.

🟩 Bard

Bard is an advanced chatbot powered by the Google Gemini model. It was launched for public use in May 2023. Similar to ChatGPT, Bard can be utilized in a wide range of applications, including customer support, content generation, as a virtual assistant etc. Bard chatbot is integrated with Google's search engine to access and provide current information also.

🟥 API

An API, which stands for “Application Programming Interface” is a set of rules and protocols for building and interacting with software applications. It enables different software systems to communicate with each other, allowing them to exchange data and functionality efficiently. A popular example of API is ChatGPT API. To summarize, an API is like a messenger that allows different apps to communicate with each other and exchange information seamlessly without having to know each other's internal workings.

🟫 OpenAI

OpenAI is a popular AI company known for its advanced GenAI models like GPT-3, GPT-4, DaLLE-3 and Whisper. Open AI was founded in December 2015 and the journey of Open AI in the Generative AI space started with the introduction of the GPT-1 model in June 2018.