University Libraries' Staff Guide to Generative AI: What is GenAI?

The video "A Timeline of Modern GenAI" by the LITE Center Professional Development Department¹ offers a comprehensive overview of the evolution of GenAI. It traces the history of GenAI research that began in the 1950s to the recent advancements in GenAI technologies like ChatGPT. It also highlights key developments and breakthroughs that have shaped the field.

The video titled “Generative AI in a Nutshell - how to survive and thrive in the age of AI” by Henrik Knibery¹, introduces generative AI, explaining its applications and how it works. It covers the benefits and challenges of generative AI, such as creating original content like text, images, audio, and video. The video also discusses the skills needed to interact with generative AI effectively, using the metaphor of having a giant brain (like Einstein) in your basement that can answer questions and take on various roles, but with some limitations and quirks. 

 

Hargrove, N.A. (2024). Image of futuristic GENAI with neurons [AI-generated image]. Microsoft Copilot.

As the field of GenAI continues to evolve, it's essential to familiarize yourself with its foundational concepts and terminology.  This list of key terms will provide a comprehensive overview of the critical components and technologies driving GenAI.
 

Key Terms

From neural networks and deep learning to Generative Adversarial Networks (GANs) and Natural Language Processing (NLP), understanding terminology will equip you with the knowledge needed to navigate and leverage the capabilities of GenAI effectively and participate in GenAI conversations with confidence.

• Artificial Intelligence: Computer systems that can perform tasks that typically require human intelligence, such as visual perception, speech recognition, decision-making, and language translation.
• Machine Learning:  A subset of AI that involves training algorithms to learn from and make predictions or decisions based on data.
• Supervised Learning (Classification):  A type of machine learning where the model is trained on labeled data, meaning the input includes examples of the correct output.
• Unsupervised Learning (Clustering):  A type of machine learning where the model is trained on unlabeled data and must find patterns and relationships in the data on its own.
• Deep Learning:  A subset of machine learning that uses neural networks with many layers (deep neural networks) to model complex patterns in data.
• Neural Networks:  Computational models inspired by the human brain, consisting of interconnected nodes (like neurons) that process data in layers.
• Embeddings:  Turn data, like words or images, into numbers that show how similar they are.  This helps the AI understand the relationships between different pieces of data.
• Token:  A unit of text (such as a word or subword) used in natural language processing.
• Prompt:  The input given to an AI model to generate a response or perform a task.
• Retrieval Augmented Generation (RAG):  A technique that combines retrieval of relevant documents with generative models to produce more accurate and informative responses.  
• Large Language Models (LLMs):  AI models that are trained on vast amounts of textual data to understand and generate human-like language.
• Foundation models:  Large-scale models trained on broad data that can be fine-tuned for specific tasks.
• Unimodal:  AI systems or models that process and analyze a single type of data, such as text, images, or audio.
• Cross-modal:  AI systems that can transfer information or learning from one modality (type of data) to another, such as using text to generate images.
• Multimodal:  AI systems that can process and understand multiple types of data simultaneously, such as combining text, images, and audio to provide more comprehensive insights.
• Hallucination:  When an AI model generates information that is not based on the input data or reality, often seen in language models.
• Bias:  Systematic errors in AI systems that can lead to unfair or inaccurate outcomes, often due to biased training data.
• Fine-tuning:  The process of taking a pre-trained model and further training it on a specific dataset to improve performance on a particular task.
• Guardrails:  Mechanisms or guidelines put in place to ensure AI systems operate safely and ethically.
• Reinforcement Learning with Human Feedback (RLHF):  A method where AI models are trained using feedback from humans to improve their performance on tasks.

Go Deeper

For a deeper dive into GenAI terms, check the Coursera article titled "Artificial Intelligence (AI) Terms: A to Z Glossary.¹"  It is an invaluable resource for anyone looking to understand the complex terminology associated with GenAI.  

¹Coursera. (n.d.). Artificial Intelligence (AI) Terms: A to Z Glossary. Coursera. Retrieved September 13, 2024, from https://www.coursera.org/articles/ai-terms