AI in Behavioural Research Repository

Artificial Intelligence (AI) is a field of science focused on building systems that can produce things like content, predictions, recommendations, or decisions to help achieve specific human goals. While AI is a subfield of computer science, the term is also often used more broadly to describe technologies that can carry out tasks that typically require human intelligence, such as recognising images, making judgments, reasoning, or making decisions.

Although tools like calculators, mobile apps, and computer programs can perform tasks that seem intelligent, they’re not usually considered AI because they don’t learn from data. In contrast, AI systems can adapt by changing how they work based on the data they receive. While saying that a machine can "learn" might sound overly human-like, it makes sense if we define learning as simply changing behaviour in response to new information. Under this broader definition, many things (not just humans) can learn.

Machine learning (ML) is a type of AI that uses data and algorithms to help systems improve their performance over time, similar to how humans learn. Machine learning is unique because it is trained on existing data and then uses what it has learned to spot patterns, make predictions, or complete tasks when it encounters new, unseen data.

A specific kind of machine learning, called deep learning, takes this further by organising the algorithms into multiple layers, which creates what is known as artificial neural networks. These networks are inspired by how the human brain works and are often behind the most realistic AI interactions, such as natural-sounding voice assistants or human-like chatbots.

Generative AI refers to advanced deep-learning models that can create high-quality content, such as text, images, or even audio, based on the data they were trained on. One of the most well-known types of generative AI is the large language model (LLM). Tools like OpenAI’s ChatGPT and Google’s Gemini fall into this category. These systems can analyse, edit, translate, and generate content that sounds natural and human-like.

LLMs work by learning from massive amounts of text data, such as websites, books, journal articles, and magazines. They use statistical methods to predict the most likely words or phrases to follow a given prompt, and they rank their responses based on how likely they are to seem correct to a human reader.

Caltech’s Science Exchange section on Artificial Intelligence has some nice accessible introductory articles covering common questions about what AI is, how it works, and what it is capable of.

Watch an event, chaired by Professor Liam Delaney, that was streamed live on October 9th 2024 from London School of Economics.  

In this public event, speakers associated with pioneering work on AI in relation to behavioural science, as part of their own research or organisational initiatives, discussed their views on how AI will change and is already changing behavioural science. 

This episode explores the next frontier of behavioural science with researchers at the forefront of AI in behavioural research. It features experts from the University of Chicago, the UAE's Behavioural Science Group, and the Max Planck Institute discussing AI applications, including super-personalised messaging, 'option C thinking,' tailored large language models for policy understanding, and the ethics of AI transparency. 

This episode explores the next frontier of behavioural science with researchers at the forefront of AI in behavioural research. It features experts from the University of Chicago, the UAE's Behavioural Science Group, and the Max Planck Institute discussing AI applications, including super-personalised messaging, 'option C thinking,' tailored large language models for policy understanding, and the ethics of AI transparency.