The following terms and concepts have been defined in the context of their use within the NSW Government.
Source: NSW Artificial Intelligence Assessment Framework, Digital NSW.
- Administrative decision making
- Administrative decisions are usually made under legislation and are directed towards a particular person (or organisation). They are different from contractual and commercial decisions, and policy and political decisions.
- Artificial intelligence (AI)
- The ability of a computer system to perform tasks that would normally require human intelligence, such as learning, reasoning and making decisions. AI encompasses various specialised domains that focus on different tasks and includes automation.
- AI system
- A machine-based system that, for explicit or implicit objectives, infers from the input it receives, how to generate outputs, such as predictions, content, recommendations or decisions that can influence physical or virtual environments. Different AI systems vary in their levels of autonomy and adaptiveness after deployment.
- Bias
- In data, this means a systematic distortion in the sampled data that compromises its representativeness. In algorithms, it describes systematic and repeatable errors in a computer system that create unfair outcomes, such as privileging one arbitrary group of users over others.
- Data governance
- Implementing a set of policies, processes, structures, roles and responsibilities to ensure that an agency’s data is managed effectively and can meet its current and future business requirements.
- Data lifecycle
- A data lifecycle illustrates the stages of data management required over time; from the time of planning and creation to the time that data is either archived or destroyed.
- Data quality
- Data quality is generally accepted as meaning ‘fitness for purpose’. It is a term used to describe a documented agreement on the representation, format and definition for data.
- Data use sensitivity
- Means risks or considerations associated with data subjects themselves, or use of data.
- Elevated risk
- Elevated risk involves systems influencing decisions with legal or similar level consequences, triggering significant actions, operating autonomously, using sensitive data, risking harm and lacking explainability.
- Generative AI
- Artificial intelligence capable of generating text, images or other media, using generative models. Generative AI models learn the patterns and structure of their input training data and then generate new data that has similar characteristics.
- Hallucination
- A hallucination or artificial hallucination is a response generated by an AI which contains false or misleading information.
- Harm
- Any adverse effects experienced by an individual (or organisation), including those which are socially, physically or financially damaging.
- Human rights
- Rights inherent to all human beings, regardless of race, sex, nationality, ethnicity, language, religion or any other status. Human rights include the right to life and liberty; freedom from slavery and torture; freedom of opinion and expression; the right to work and education; and many more. Everyone is entitled to these rights, without discrimination.
- Large language model (LLM)
- A specialised type of artificial intelligence that has been trained on vast amounts of text to understand existing content and generate original content.
- Operational AI
- Systems that have a real-world effect. The purpose is to generate an action, either prompting a human to act, or the system acting by itself. Operational uses of AI often work in real time (or near real time) using a live environment for their source data.
- Responsible officer
- The officer who is responsible for the use of the AI insights or decisions; the outcomes from the project; the technical performance of the AI system; and data governance.
Reversible harm
Means an adverse effect that can be reversed with some level of effort, cost and time.
Secondary harm
Means any adverse effects experienced by an individual (or organisation) not directly engaged with the AI system, or a subsequent harm identified after an initial harm is experienced by an individual (or organisation) engaged with the AI system.
Significant harm
Always context specific, a harm which leads to significant concerns.
Additional terms not defined by the NSW Government
Explainability
Property of an AI system to express important factors influencing the AI system results in a way that humans can understand.
Fairness
Treatment, behaviour or outcomes that respect established facts, beliefs and norms and are not determined or affected by favouritism or unjust discrimination (ISO TR24368).
Unfairness
Unjustified differential treatment that preferentially benefits certain groups more than others.
Reliability
An AI system's ability to consistently perform its intended functions accurately and dependably over time and under various conditions. Minimising errors and ensuring predictable, trustworthy results.