Key Points

Introduction

Artificial Intelligence is a broad field concerned with systems that perform tasks associated with human intelligence.
Machine learning is a subset of AI that learns from data.
Deep learning is a subset of machine learning based on multi-layered neural networks.
Large language models are a specific type of deep learning model focused on language.
The recent growth of AI has been driven by data availability, increased computing power, and algorithmic advances.

Machine learning systems learn patterns from data rather than following rules.
Training and test sets help us assess whether a model generalises to new data.
Interpretable models make their reasoning transparent whereas black box models do not.
Traditional statistical methods are often more appropriate than machine learning when the goal is explanation rather than prediction, particularly with small datasets.
The quality and representativeness of training data strongly influence model performance and fairness.

Artificial neural networks consist of layers of weighted computational units inspired by biological neurons.
‘Deep’ refers to having multiple hidden layers that learn increasingly abstract representations.
Training involves making predictions, measuring error, and adjusting weights using backpropagation.
Deep learning excels at complex pattern recognition tasks such as image, audio, and text analysis.
Large models require extensive data and computing resources to train effectively.

LLMs are deep learning models trained on massive text datasets to predict the next word, from which broad language capabilities emerge.
The Transformer architecture, and its attention mechanism, is the foundation of all major modern LLMs.
Pre-training builds general language knowledge; fine-tuning specialises a model for particular tasks or behaviours.
LLMs hallucinate — they generate confident but factually incorrect content — because they are optimised for coherent text, not verified truth.
LLMs have a knowledge cutoff date and are unaware of more recent events unless equipped with external search tools.
Outputs are probabilistic: the same prompt can produce different responses, with implications for research reproducibility.

AI techniques are being applied across research disciplines, from text analysis and image classification to code generation and structured data modelling.
Before adopting any AI tool, ask: what was it trained on? Has it been validated? Can results be reproduced? Can outputs be explained? What are the failure modes?
AI models reflect the biases in their training data.
Transparency in methods is essential: report which tools were used, at what version, for what purpose, and how outputs were validated.
Privacy and data governance must be considered before inputting any sensitive or personal data into an AI tool.
Authorship, attribution, and environmental cost are emerging ethical considerations that researchers should engage with actively.
Consider the impacts on human intelligence when outsourcing tasks to AI
Developing AI literacy is an ongoing practice: follow institutional guidance, read model documentation, and engage with methodological debates in your own field.