Fundamentals of AI and Machine Learning

This chapter builds the conceptual and practical foundation required to apply artificial intelligence (AI) and machine learning (ML) to chemical problems, moving from core definitions to modern model families and their deployment in real workflows. It opens by positioning AI and ML as computational approaches to intelligence—systems that learn patterns from data—and then introduces the principal paradigms of ML, clarifying how supervised learning, unsupervised learning, and related formulations map onto common chemistry tasks such as property prediction, classification of outcomes, clustering of chemical space, and decision-making under uncertainty. This chapter provides an accessible overview of key algorithms that dominate applied chemical ML, emphasizing what they assume about data, what they optimize, and the practical trade-offs between interpretability, flexibility, and data efficiency. It then develops the basis for neural networks and deep learning, explains the essential components and intuitions behind representation learning, and surveys mainstream deep-learning architectures that have become central to modern chemical AI. Special attention is given to training and evaluation—how models are optimized in practice, how overfitting emerges, and how validation design, metrics, and error analysis determine whether performance is meaningful for prospective chemical use. Building on these fundamentals, the chapter introduces large language models (LLMs) in the context of chemistry, outlining what LLMs are, why they have become influential, and how they can be used as interfaces for chemical knowledge and reasoning. It discusses the role and key applications of LLMs in chemistry and includes a focused case study on predicting molecular properties with LLM-based approaches, followed by an honest discussion of challenges and future directions, including reliability, controllability, and domain grounding. This chapter then connects methods to practice through a set of representative applications in chemistry—predictive modeling, reaction prediction and mechanism exploration, and molecular design and optimization—illustrating how algorithm choices and representations shape scientific conclusions. Finally, it expands from modeling to execution by addressing integration with automation and robotics, explaining why robotic systems matter, how AI-driven automation enables machine-guided experimentation, and what real implementations look like, before closing with ethical and regulatory considerations that frame responsible development, compliance, and safe deployment in AI-driven chemistry.