AI terms explained: a practical workplace glossary

Artificial intelligence is becoming part of everyday work. You may see it in writing tools, customer service systems, recruitment platforms, data analysis, cybersecurity and workplace software. It's all throughout social media, on our phones, and in many apps. Understanding the language helps you ask better questions, use the technology more confidently and recognise when human judgement still matters.

This plain-English glossary explains the AI terms employees and job seekers are most likely to encounter. You do not need a technical background. Start with the essentials, then use the practical examples to connect each term with real work.

The essential AI terms to know at work

1. Artificial intelligence (AI)

A broad term for computer systems that perform tasks associated with human intelligence, such as recognising patterns, understanding language, making predictions or generating content. AI is an umbrella term, not one single tool.

2. Algorithm

A set of instructions or rules a computer or system follows to complete a task or solve a problem. Algorithms can be simple, or they can help power complex systems such as search results, social media feeds, recommendations and fraud detection.

3. Machine learning

A type of AI that learns patterns from data and uses those patterns to make predictions or decisions. For example, a system might learn to identify likely spam emails after analysing many labelled examples.

4. Deep learning

A form of machine learning that uses layered neural networks to find complex patterns. It supports applications such as speech recognition, computer vision and some generative AI systems.

5. Neural network

A computing model made of connected processing units, often called nodes or neurons. Neural networks learn relationships within data and are loosely inspired by how biological brains process information.

6. Generative AI

AI that creates new content, including text, images, audio, video or code, based on patterns learned from existing data. A generative AI response can be useful without being automatically accurate.

7. Large language model (LLM)

A model trained on very large collections of text so it can process and generate language. LLMs power many conversational AI tools, but they do not understand information in the same way a person does.

8. Natural language processing (NLP)

Technology that helps computers work with human language. It can support translation, speech recognition, sentiment analysis, document summarisation and conversational tools.

9. Chatbot

Software designed to interact with people through written or spoken conversation. Some chatbots follow fixed rules, while others use generative AI to create more flexible responses.

10. Prompt

The instruction, question or context you give an AI system. A useful prompt normally states the task, relevant background, intended audience, constraints and output format. You may have heard someone say 'the better the prompt, the better the response', this helps ensure we get a better answer from AI, not just something generic.

Terms you will hear as workplace AI becomes more capable

11. AI assistant

An AI-powered tool designed to help a person complete tasks such as drafting, summarising, researching or organising information. Unlike more autonomous AI agents, an AI assistant generally works in response to a person’s instructions, with the person remaining in control of the task.

12. AI agent

An AI-enabled system designed to pursue a goal and take a sequence of actions, sometimes using other tools or data sources. Unlike a standard chatbot that mainly responds to individual prompts, an AI agent may plan and complete several steps towards a task.

13. Agentic AI

A broader term for AI systems designed to operate with greater independence. Agentic AI can plan, make decisions, use tools and adapt its actions while working towards a goal. An AI agent is an example of how agentic AI can be put into practice.

14. Multimodal AI

AI that can work across more than one type of information, such as text, images, audio and video. A multimodal tool might analyse a chart and then explain it in writing.

15. Automation

The use of technology to complete tasks with reduced manual effort. Automation is not always AI. A fixed workflow can be automated without learning or generating new content.

16. Application programming interface (API)

A defined way for software systems to communicate. An API can allow an approved workplace application to send information to an AI service and receive a result.

17. Retrieval-augmented generation (RAG)

A method that allows a generative AI system to reference selected information before producing an answer. For example, it might retrieve relevant content from an approved knowledge base or set of documents. This can help ground the response in current or organisation-specific information, but the sources and final answer should still be checked.

18. Fine-tuning

Additional training that adapts an existing model for a narrower task, style or domain. Fine-tuning does not automatically remove bias, errors or security risks.

19. Training data

The information used to teach an AI model patterns. The quality, coverage, permissions and limitations of that data can affect how the system performs.

20. Inference

The stage when a trained model processes new input and produces an output. When you ask a generative AI tool a question and receive an answer, the model is performing inference.

Terms that help you assess AI output responsibly

21. Hallucination

A confident-sounding AI output that is incorrect, unsupported or invented. Hallucinations are why important facts, quotations, calculations, references and decisions need to be verified against reliable sources.

22. Bias

A systematic unfair tendency in data, design or outcomes. Bias can enter through historical data, incomplete representation, design choices or the way a tool is used. It can affect people differently and should be actively tested and managed.

23. Explainable AI

Approaches that help people understand how an AI system reached a result or which factors influenced it. The right level of explanation depends on the decision and its impact.

24. Human-in-the-loop

A process in which a person reviews, guides or approves an AI-assisted task. Human oversight is especially important when an output could affect employment, safety, privacy, finances or someone’s access to an opportunity.

25. AI governance

The policies, responsibilities and controls used to manage AI. Governance can cover approved tools, data use, privacy, security, risk assessment, recordkeeping, monitoring and accountability.

26. Transparency

Being open about when and how AI is used, particularly when people may be affected by an AI-supported interaction or decision. Transparency should be meaningful, not simply a technical disclaimer.

27. Model drift

A decline or change in performance as real-world conditions and data change over time. An AI system that worked well at launch may need ongoing monitoring and review.

28. Deepfake

Synthetic or manipulated audio, video or images designed to make something appear real. Deepfakes can be used for fraud, misinformation or impersonation, so unexpected requests should be verified through a trusted channel.

29. Data privacy

The responsible handling of personal information. Before entering workplace, customer or candidate information into an AI tool, check whether the tool is approved and follow your organisation’s privacy and information-security requirements.

30. Responsible AI

The practice of designing, developing and using AI in ways that are trustworthy and appropriate for the context. In New Zealand, MBIE provides voluntary guidance to help businesses manage AI responsibly, including considerations around governance, privacy, security, transparency, risk and accountability.

How to use AI more confidently at work

Knowing the terminology is useful. Knowing what to do with it is better. Before using an AI tool for a workplace task, pause and apply these five checks:

• Purpose: Be clear about the problem you are trying to solve and whether AI is the right tool.
• Permission: Use approved systems and understand what information you are allowed to enter.
• Proof: Check important claims, calculations and references against reliable sources.
• People: Consider who could be affected by the output and where human judgement is required.
• Protection: Follow privacy, confidentiality, cybersecurity and recordkeeping requirements.

If the result could influence hiring, performance, safety, legal obligations or access to an opportunity, involve the right person and apply a higher level of review. AI can support good judgement. It should not replace accountability.

Frequently asked questions about AI terminology

What is the difference between AI and generative AI?

AI is the broad category. Generative AI is one type of AI that creates new content from patterns in its training data. Other AI systems may classify, predict, recommend or detect without generating content.

What is the difference between an AI agent and a chatbot?

A chatbot is designed mainly for conversation. An AI agent may use conversation as an interface, but it is designed to pursue a goal and take multiple actions. Some tools combine both.

Can I trust an AI-generated answer?

Treat it as a draft or starting point unless it has been checked. The level of verification should match the risk. A brainstorming suggestion needs less scrutiny than information used for a hiring, safety, financial or legal decision.

Should I put confidential information into an AI tool?

Only if the system is approved for that information and your organisation’s rules allow it. If you are unsure, do not enter personal, confidential or commercially sensitive material. Ask the appropriate privacy, security or technology contact first.

Will understanding AI terminology help my career?

It can help you participate in workplace conversations, ask more informed questions and describe how you use technology responsibly. The strongest skill is not memorising jargon. It is knowing what a tool can do, where it can fail and when people need to step in.

Keep building your AI and career knowledge

Explore how generative AI can support a job application in our article on AI-powered CV writing. Then browse more Adecco New Zealand career insights for practical guidance on work, skills and your next move.

For New Zealand organisations, the Responsible AI Guidance for Businesses is a useful starting point for trustworthy AI adoption.