AI, or artificial intelligence, is the ability of machines to imitate facets of human intelligence. Though many think that AI is a very recent phenomenon, the official beginning of AI came in the form of a summer program at US Ivy League university Dartmouth College in 1955, with the hypothesis that:

‘every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it’.

Professor Stuart Russell, a pioneer in human-compatible AI, identifies the following key aspects of AI models: (1) the ability to choose an action ‘by looking ahead and considering the outcomes of different possible action sequences’; (2) the ability to carry out logical reasoning based on definite knowledge; and (3) the ability to learn from examples.

Generative AI is a specific type of AI technology which can produce content; this content can take numerous forms, but the most well-known is perhaps natural language – i.e. speech and text. Generative language models work by identifying patterns in language in so-called ‘training data’; then, given some text, they can use these patterns to predict what would come next (for example, predicting the next word in a given sequence of text).

Typically trained on massive amounts of data, these models can become extremely powerful – one common example is ChatGPT (Chat Generative Pre-trained Transformer), a text-to-text generative language model that can create virtually any new text-based content based on what it has learned from existing content (the data it has been trained on).