Essentials
About Generative AI
For decades, the concept of “machine learning” has been incorporated into a range of applications, including optical character recognition, fraud detection in financial transaction data, medical image analysis, video surveillance systems, and voice recognition in “assistants” like Siri and Alexa. Yet while remarkable and often highly sophisticated, these traditional AI technologies focus on analyzing existing data to identify patterns and make predictions, mimicking human work that has already been done.
Generative AI uses patterns it detects from vast data on the internet and other public sources to create entirely new content. Essentially, it is a sophisticated predictor for text content, image, code and audio, chaining together the most likely next word, pixel, line of code or sound wave to mimic humans and generate in a human-like way. However, it is important to note that while much of its output can be incredibly useful, it can make surprising and sometimes significant errors. In order to prevent harm, inaccuracy, and bias that can arise from these errors, one should always maintain at a minimum, the role of “expert editor,” for any AI content, an approach commonly known as “human in the loop.”
Key AI Terms and Definitions
As AI becomes more integrated into education, research, and professional work, understanding essential AI-related terms can help you navigate and leverage these technologies effectively. Here are some fundamental terms and their meanings:
| AI | Artificial Intelligence (AI) is technology that enables computers and machines to simulate human learning, comprehension, problem solving, decision making, creativity and autonomy (definition by IBM) |
| AI Chatbot | An AI Chatbot is software the simulates conversation with human end-users (definition by IBM) |
| AI Bias | AI bias, also called machine learning bias or algorithm bias, refers to the occurrence of biased results due to human biases that skew the original training data or AI algorithm—leading to distorted outputs and potentially harmful outcomes. (definition by IBM) |
| Deepfake | A deepfake is a fake but convincing image, video, or audio created with AI to intentionally mislead or manipulate |
| Generative AI | Generative AI is a type of AI that is specifically intended to create new content such as text, images, audio and video output, based on data it has been trained on. |
| Hallucination | A hallucination is an inaccurate, nonsensical output generated by AI. |
| Human-in-the-loop | Human-in-the-Loop is a design approach where humans are actively involved in the training and testing of AI systems. |
| Large Language Model (LLM) | Large language models are AI systems capable of understanding and generating human language by processing vast amounts of text data. (definition by IBM) |
| Machine Learning | Machine learning (ML) teaches AI machines how to perform specific tasks and provide accurate results by identifying patterns |
| Output | Output is new data created by generative AI in response to a prompt |
| Prompt | A prompt is a question, statement, or other input that is given to an AI model to generate a specific response |
| Prompt Engineering | Prompt engineering is the process of designing and revising inputs for AI models to create the desired outputs. It involves carefully crafting questions or statements to provide context, instructions, and examples to help the AI understand the desired output. |
| Training Data | Training data is the set of information used to teach AI models to generate accurate predictions, decisions, and output. |
Quick Start Resources
We invite you to explore this site and its many resources, but the below sections may provide a good starting point.
Suggested Reading
- Co-intelligence : living and working with AI, Ethan Mollick, 2024
An illuminating, entertaining and fascinating read, Ethan Mollick’s “Co-Intelligence” presents a timely and pragmatic examination of the evolving relationship between humans and generative artificial intelligence, particularly relevant for an academic audience. - A Generative AI Primer, from the UK’s National Centre for AI at JISC.
- Generative AI could be the right tool for a given task, but not always. Some may enjoy reading 15 Times to Use AI and 5 Not To, by Ethan Mollick.