Название: Artificial Intelligence for Learning: Using AI and Generative AI to Support Learner Development, 2nd Edition
Автор: Donald Clark
Издательство: Kogan Page
Год: 2024
Страниц: 313
Язык: английский
Формат: pdf (true), epub
Размер: 22.7 MB

With Artificial Intelligence (AI) creating huge opportunities for learning and employee development, how can learning professionals best implement the use of AI into their environment? Artificial Intelligence for Learning is the essential guide for learning professionals who want to understand how to use AI to improve all aspects of learning in organizations. This new edition debunks the myths and misconceptions around AI, discusses the learning theory behind generative AI and gives strategic and practical advice on how AI can be used. This book also includes specific guidance on how AI can provide learning support, chatbot functionality and content, as well as ideas on ethics and personalization. This book is necessary reading for all learning practitioners needing to understand AI and what it means in practice.

AI in one sense means doing what humans do when they learn. The AI field is thick with references to ‘learning’, the most common being machine, deep and reinforcement learning. Machine Learning uses algorithms and statistical models and applies patterns and inferences to perform tasks through experience. This is analogous to traditional learning through exposure to taught experiences. Deep Learning is a Machine Learning technique that uses layered neural networks with large datasets, supervised, semi-supervised or unsupervised, to perform tasks. This is more like learning from the real world to gain competence and solve actual problems. Reinforcement learning operates on maximizing reward by exposure to existing knowledge and exploring new knowledge. This is more like deliberate practice to get things right.

Cognitive AI enables search across the web and access to educational resources such as knowledge on Wikipedia and ‘how to’ videos on YouTube. It can make connections, analyse data, use data to predict, prescribe, translate and transcribe. There’s the detection and help with those who have disabilities and learning issues. It can now also create, compose new music, paintings, sculpture. In learning this is by far the most significant area, as text and images, which can now be generated, play a significant role in learning.

As LLMs emerged, it became obvious that something extraordinary had happened in AI. The emergent qualities in terms of generating well-structured, grammatically correct text that seemed almost human, amazed the world. It could translate and generate answers in depth over a breadth of subjects. The same happened with images, the quality improving day by day. As generative AI became multimodal, with speech input and output, as well as interpretation of uploaded images, it gained in functionality. Then user created chatbots were launched, allowing anyone to create a generative pre-trained transformer (GPT) or personal chatbot. All of this within one year. At the heart of it all were LLMs trained using gargantuan amounts of data. There seemed to be something in the model and data, in language itself that enabled this miracle.