ICAEW.com works better with JavaScript enabled.

Discover the key technical terms associated with generative AI and their meaning.


– refer to Application Programme Interfaces that enable the flow of information between applications. An API would allow an AI chatbot to connect to a LLM by a third-party service provider.

Artificial neural networks

- are computer programs or algorithms that simulate the behaviour of neurons in the brain. ANNs are a subset of machine learning and have enabled deep learning progress by enabling training at multi-layer neural network levels. 

Read more:

Deep learning

– refers to a process of learning for computers. Computer models are exposed to large volumes data which is compared to expected outcomes to continuously refine their "thinking” and enable learning. 

Read more:


– refers to an ultra-realistic video or audio of a person in which they are made to say or do something that is fake. Wider availability of amplified processing power and larger data sets means that creating a deepfake is remarkably straightforward.  

Read more:

Foundation models

- are large machine learning models trained on large volumes of data at scale. These models are capable of a range of general tasks. OpenAI’s GPT is an example of a foundation model. 

Read more:


GitHub is a code hosting platform allowing developers to store and manage their code for version control and collaboration. 

Read more:


- stands for Generative Pre-trained Transformer. GPT is an example of a large language model created by OpenAI that has been trained on a large volume of data that underpins it’s chatbot- ChatGPT. 


– refer to Large Language Models that have been trained on a large volume of text-based data. Sources of data usually differ depending on the LLM. The algorithms underpinning the LLMs are able to analyse the data and probabilities so when the model is given a prompt, for example in the form of a question, the model is able to generate an answer. 

Read more:

Public model

- refers to publicly available AI models trained on a large volume of data from various sources. OpenAI’s GPT is an example of a public model. Providers of public models control how input information is stored and incorporated into future versions of the models.  

Read more:

Synapse-style relationships

- Human brains learn from experience through synaptic plasticity. Deep learning models that are trained with large volumes of data and new algorithms learn in a similar way. These models are based on human synapse-style relationships. 
Read more:

You may also be interested in

Shape the future slogan banner
Annual Conference: Technology

Technology is rapidly evolving and transforming the way we work. ICAEW's Annual Conference 2023 focuses on the need for accountants to adapt and stay up to date.

Artificial intelligence
Artificial intelligence

Discover more about the impact of artificial intelligence and the opportunities it presents for the accountancy profession. Access articles, reports and webinars from ICAEW and resources from tech experts.

Browse resources
ICAEW Community

Do you use Excel in your organisation? Are you using it to its maximum potential? Develop your skills and minimise spreadsheet risk with our Excel resources. Join the Excel Community.