What is AI? Artificial intelligence explained in plain terms
Short answer
AI, or artificial intelligence, is software that produces outputs such as predictions, content, recommendations or decisions from the input it receives, instead of following a rule written for every case. Most AI in use today is narrow: built for one kind of task, such as ranking a feed, translating text or flagging a risky payment. Artificial general intelligence is a research goal, not a product.
On this page
What AI means, by the definitions that count
There is no single definition of AI, but the two most cited institutional ones agree on the core.
The Organisation for Economic Co-operation and Development (OECD) updated its definition on 29 November 2023: an AI system is a machine-based system that, for a given set of human-defined 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, and different systems operate with varying levels of autonomy and adaptiveness after deployment (OECD.AI).
The US National Institute of Standards and Technology (NIST) uses closely related wording in its AI Risk Management Framework 1.0, published in January 2023. It refers to an AI system as “an engineered or machine-based system that can, for a given set of objectives, generate outputs such as predictions, recommendations, or decisions influencing real or virtual environments,” and notes that AI systems “are designed to operate with varying levels of autonomy” (NIST AI 100-1).
In both, the key word is “infers.” An ordinary program runs the rules a person wrote. An AI system works out how to produce its output from the input, based on patterns it was built or trained to recognise.
How AI works in three ideas
Data
AI starts with examples. A fraud model sees many past transactions, a translation model sees many pairs of sentences, and a language model sees large amounts of text. The quality and coverage of the examples limit what the system can do.
Models
A model is the mathematical structure that holds what was learned from the data. Google’s machine learning glossary defines machine learning as “a program or system that trains a model from input data” (Google for Developers, Machine Learning Glossary).
Training and use
Training is the process of fitting the model to the data. Use, sometimes called inference, is when the trained model gets new input and produces an output. Businesses rarely train models. They usually call a model that a provider has already trained, and the work is in giving it the right information, setting limits and checking results.
Types of AI
| Term | Plain meaning | Example |
|---|---|---|
| Artificial intelligence | The broad field and the systems it produces | Any item below |
| Machine learning | AI that learns patterns from data instead of following hand-written rules | A model that flags unusual payments |
| Narrow AI | AI built for one type of task or a small set of tasks | A translation tool, a feed-ranking system |
| Generative AI | AI that produces new content in response to a prompt | A tool that drafts a reply or summarises a document |
| Large language model | A generative model trained on text, behind most chat assistants | The model inside a chat assistant |
| Artificial general intelligence (AGI) | A hypothetical system with broad, human-level ability across tasks | None exists as a product |
Swipe the table sideways to see all columns.
Narrow AI
“Narrow AI” is a common informal term, not a formal standard, for AI designed for a specific job. Chat assistants, translation tools and fraud models are all examples. Each is built for a defined set of jobs, even when the set is wide.
Generative AI
Generative AI produces new text, images, audio or code. In business it is mostly used to draft, summarise and extract information. Because it generates fluent output whether or not it is true, wrong answers are a known risk, called hallucination.
Artificial general intelligence as a research goal
Definitions of AGI vary. Google’s glossary calls it “a non-human mechanism that demonstrates a broad range of problem solving, creativity, and adaptability.” A group of eight researchers wrote in 2023 that they analysed “existing definitions of AGI” and proposed a framework with levels of performance and generality (Morris et al., arXiv, 4 November 2023, revised 24 September 2025). That an academic paper had to propose a way to classify AGI shows that the term has no settled definition. For a business reader, AGI is a research goal and not a product to buy or plan around.
Machine learning and where it sits
Machine learning is a subfield of AI. Google’s glossary states: “Formally, machine learning is a sub-field of artificial intelligence.” Most modern AI, including generative AI, is built with machine learning.
Examples people in the Philippines already use
Each example below points to something the company or regulator has published. Where no public source was found, the description is general.
- Ride and delivery apps. Grab said in October 2023 that it “previously” maintained over 200 unique machine learning models to calculate estimated arrival times (Inside Grab, 6 October 2023). The article is about the company’s regional app, not Philippine data specifically.
- Bank and e-wallet fraud checks. The Bangko Sentral ng Pilipinas (BSP) Circular No. 1213, Series of 2025, approved 22 May 2025, requires supervised institutions to implement automated and real-time fraud monitoring and detection systems, and states that they “shall leverage a combination of rule-based approaches, machine learning algorithms, and other technologies” (BSP Circular No. 1213).
- Translation. Google’s Cloud Translation language list includes Filipino, Cebuano and Iloko among its supported languages (Google Cloud, Language support, read 12 October 2026).
- Feed and shopping recommendations. Meta says Facebook uses “state-of-the-art machine learning systems to rank content” in Feed (Meta Transparency Center, updated 11 June 2025).
- Chat assistants. Google describes its Gemini apps as giving “direct access to Google AI” for writing, planning and learning (Google Gemini Apps Help). The AI assistants explainer covers the types.
AI at work
- Writing and summarising. Drafting a reply, shortening a long email thread, preparing a first version of a notice for a person to edit.
- Answering customers. A chatbot answering Messenger and Viber questions about hours, prices and bookings, with a handover to staff.
- Sorting documents. Reading invoices, receipts or forms and pulling out the supplier, date and total. This is a common step in AI automation.
- Forecasting. Predicting demand or stock needs from past sales. The input is numbers, and the forecast can be checked later against what happened.
The AI in the Philippines report shows how widely firms use AI. The most recent firm-level survey figure found is 3.02 percent of establishments in 2021, with higher rates in BPO (7.19 percent) and ICT (5.94 percent). That survey is five years old, and a current official figure was not found.
What AI does well and where people stay in charge
AI does well when there are many similar inputs, patterns that people cannot write as rules, and an output that someone can check. It handles large volumes consistently and reads language that does not follow a fixed format.
It does less well when the facts are missing from what it was given, when the situation is new, or when a wrong answer is costly. It cannot take responsibility. A person or business stays accountable for what a system tells customers, for how personal data is handled under the Data Privacy Act of 2012, and for any decision about health, money or legal rights. Keeping a person in the review step, as the human in the loop entry explains, is how most businesses manage this.
Terms that follow from here
- Automation: software doing a defined task without a person at each step.
- AI agents: AI systems that choose their own steps and tools to reach a goal.
- Chatbots: programs people talk to in a chat window.
- Generative AI: AI that produces new content.
- Conversational AI: AI that understands and produces human conversation in text or speech.
Frequently asked questions
What is AI in simple words?
AI is software that works out its output from the input it receives, such as a prediction, a piece of text or a recommendation, instead of following a rule written for each case.
What are the main types of AI?
Narrow AI is built for specific tasks and describes nearly all AI in use today. Generative AI produces new content. Artificial general intelligence is a research goal for systems with broad human-level ability. Machine learning is the method behind most of it.
What is the difference between AI and machine learning?
AI is the broad field. Machine learning is a part of it in which a system learns patterns from data. Most current AI uses machine learning.
What is artificial general intelligence?
AGI is a hypothetical AI with broad ability across tasks. Definitions differ between organisations, and no AGI product exists.
Where is AI used in daily life in the Philippines?
Published examples include fraud monitoring that BSP rules require of banks and e-wallets, translation tools that cover Filipino, Cebuano and Iloko, feed ranking on Facebook, and ride and delivery apps that estimate arrival times.
Sources
All sources were read on 12 Oct 2026.
- OECD.AI, Updated OECD definition of an AI system, 29 November 2023.
- NIST, Artificial Intelligence Risk Management Framework (AI RMF 1.0), NIST AI 100-1, January 2023.
- Google for Developers, Machine Learning Glossary.
- Morris et al., Levels of AGI for Operationalizing Progress on the Path to AGI, arXiv, 4 November 2023 (v5, 24 September 2025).
- Grab, How we got more accurate estimated time-of-arrivals in the app, 6 October 2023.
- Bangko Sentral ng Pilipinas, Circular No. 1213, Series of 2025, approved 22 May 2025.
- Google Cloud, Language support, Cloud Translation.
- Meta, Ranking and content, Transparency Center, updated 11 June 2025.
- Google, Gemini Apps Help.
- Philippine Institute for Development Studies, Discussion Paper 2024-35, December 2024 (via the AI in the Philippines report).