Glossary of AI terms

Accessibility is considering the needs of all potential users from the outset, engaging with individuals who have disabilities during the design process in order to create solutions that are genuinely usable by everyone.

It also includes: Assistive Technologies, screen readers, voice recognition software and alternative input devices.

Bias in AI models typically arises from two sources: the design of models themselves and the training data they use.

Models can sometimes reflect the assumptions of the developers coding them, which causes them to favour certain outcomes.

Additionally, AI bias can develop due to the data used to train the AI.

Governance involves steering responsible development, deployment, and use of AI technologies throughout their lifecycle, by creating and implementing a range of tools such as voluntary guidelines, policies, regulations, and laws, amongst others.

An AI system lifecycle typically involves several phases that include:

• to plan and design
• collect and process data
• build model(s) and/or adapt existing model(s) to specific tasks
• test, evaluate, verify, and validate
• make available for use/deploy
• operate and monitor
• retire/decommission.

These phases often take place in an iterative manner and are not necessarily sequential. The decision to retire an AI system from operation may occur at any point during the operation and monitoring phase.

Recommendation of the Council on Artificial Intelligence - OECD

An AI system is 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.

What is AI? Can you make a clear distinction between AI and non-AI systems? — OECD

Chatbot is a digital tool or software application designed to simulate conversation with users (primarily via text or synthesised speech). Some operate on predefined responses but advanced versions integrating Gen AI provide more dynamic and responsive interactions with users.

Cybersecurity involves measures to protect systems, data, and devices from unauthorised access, and ensuring the confidentiality, integrity, and availability of information.

Data can be defined as any information in a form capable of being communicated, analysed, or processed (whether by an individual or by computer or other automated means).

Data is useful when it can be communicated easily and analysed to gain insights. Data’s value stems from its use, re-use, and re-purposing, particularly in large volumes. To properly realise this value, data must be accurate, reliable, and free from bias.

Data ethics refers to the study and practice of ethical issues related to data, including its generation, recording, processing and distribution, and use. It encompasses principles and standards that guide the responsible and fair handling of data to ensure the rights and privacy of individuals are protected.

Key principles of data ethics:

• Privacy: Ensuring that personal information is collected, stored, and used in ways that protect individuals’ privacy and comply with legal requirements.
• Transparency: Being open about how the data is collected, used, and shared.
• Consent: Obtaining informed consent from individuals before collecting their data.
• Security: Protecting data from unauthorised access, breaches, or cyberattacks to maintain its confidentiality, integrity, and availability.
• Fairness: Ensuring data practices do not result in discrimination or bias and that data will be used in ways that are both equitable and just.
• Accountability: Holding organisations and individuals accountable for their data practices and ensuring there are mechanism in place to address any issues if they arise.

OECD Good Practice Principles for Data Ethics in the Public Sector — OECD

Data transparency is providing clear and accessible information about the data used in AI systems. This includes understanding where the data comes from, how it has been collected, processed, and used, and making processes more open and understandable to stakeholders.

Deep learning is a more specialised machine learning technique in which more complex layers of data and neural networks are used to process data and make decisions.

Explainability means enabling people affected by the outcome of an AI system to understand how it was arrived at. This entails providing easy-to-understand information to people affected by an AI system’s outcome that can enable those adversely affected to challenge the outcome, notably – to the extent practicable – the factors and logic that led to an outcome. Notwithstanding, explainability can be achieved in different ways depending on the context (such as, the significance of the outcomes).

Transparency and explainability — OECD.AI

Generative AI is a type of AI system that can create or generate new content such as text, images, video and music based off models and patterns detected in existing datasets. (OECD)

The OECD defines hallucinations as when GenAI systems create incorrect yet convincing outputs.

Generative AI: the risks and the unknowns — OECD.AI

Human in the loop refers to the involvement of human oversight and decision-making in the processes that involve AI and automated systems. This approach allow for critical decisions, especially those impacting individuals, to be reviewed, verified, and influenced by human judgement and expertise.

Indigenous data refers to data that is related to Indigenous peoples, their territories, resources, cultures, languages, and knowledge systems. Data is a critical part of indigenous identity and sovereignty, encompassing a wide range of information from traditional ecological knowledge to health and demographic data.

See also Māori Data Governance.

Indigenous data sovereignty thematic area narrative in English, Arabic, French, Portuguese and Spanish — Global Index on Responsible AI

Large language models, also known as LLMs, are very large deep learning models that are pre-trained on vast amounts of data.

AI language models: Technological, socio-economic and policy considerations — OECD

Machine learning (ML) is a type of artificial intelligence that allows machines to learn from data without being explicitly programmed. It does this by optimising model parameters (that is, internal variables) through calculations, such that the model’s behaviour reflects the data or experience. The learning algorithm then continuously updates the parameter values as learning progresses, enabling the ML model to learn and make predictions or decisions.

Māori data governance refers to the principles and practices that ensure Māori data is collected, managed, and used in a way that respects Māori values, rights, and interests.

Key aspects of Māori data governance include:

• Data Sovereignty – ensuring Māori data is subject to Māori governance and control.
• Ethical use, promoting the ethical use of data to enhance wellbeing of Māori people, language, and culture.
• Advocating for Māori involvement in the governance of data repositories and decision-making processes.
• Safeguarding the quality and integrity of Māori data.

Te Mana Raraunga has developed resources to better understand and to support the principles.

Te Mana Raraunga

Misinformation refers to false or inaccurate information that is spread regardless of an intent to deceive. Unlike disinformation, which is deliberately misleading, misinformation is often shared without malicious intent.

Predictive AI (or predictive analytics) Involves using data, statistical algorithms, and machine learning techniques to identify the likelihood of future outcomes based on historical data.

Risk is the likelihood of adverse impact on people, the environment and society.

As defined in the OECD Recommendation on AI: Persons or groups, or their legitimate representatives, who have rights or interests that are or could be affected by adverse impacts associated with the enterprise’s operations, products, or services.

These can include users of the AI system, civil society, workers’ representatives, service providers, and other enterprises. Anyone involved in or affected by relevant systems.

Recommendation of the Council on Artificial Intelligence — OECD

Making the operation and decision-making processes of AI systems clear and understandable to users and stakeholders. Key components of transparency are:

• Openness: Clearly communicating the purpose and capabilities of an AI system. This includes explaining what the system is designed to do and any limitations it may have.
• Explainability: Providing understandable explanations of how the AI system reaches it decisions.
• Accountability: Ensuring that there is a mechanism for tracking and verifying decisions made by the AI. This can include maintaining logs, version control and audit trails
• Data Transparency: Disclosing what data is used to train and operate the AI system, including its sources and how it is processed.

Transparency and explainability — OECD