# Artificial intelligence fundamentals

> CIE A-Level Computer Science · 9618
> Source: https://www.owlsprep.com/study/cie-9618-u13-artificial-intelligence-fundamentals/

This sub-topic covers core definitions and classifications of artificial intelligence (AI) for CIE A-Level 9618. You will learn key concepts for exam questions, including AI types, intelligent agents, and real-world AI applications.

**Prerequisites:** [Basic computational problem solving fundamentals](https://www.owlsprep.com/study/cie-9618-u13-problem-solving-fundamentals/)

## Learning objectives

- Distinguish between strong AI and weak (narrow) AI
- Explain core characteristics of intelligent AI systems
- Identify and describe common real-world applications of AI

## Classification of AI: Strong vs Weak AI

**AI Classification** — AI is split into two broad categories based on the scope of its capabilities, for assessment purposes.

*Example:* Almost all commercial AI systems in use today are classified as weak AI.

The distinction between strong and weak AI is a frequently tested topic, usually for 2-4 marks. Classification depends on how generalizable the AI's capabilities are, not just how well it performs a single task.

**Worked example:** A social media recommendation system that suggests posts for users is classified as AI. State whether this is strong or weak AI, and justify your answer.

1. Recall the core definitions: weak AI is narrow, designed for a limited set of specific tasks, while strong AI has general human-like intelligence across all domains.
2. Analyze the recommendation system: it is built only to perform the specific task of suggesting content, and cannot generalize to unrelated intellectual tasks the way a human can.
3. Conclusion: The recommendation system is weak (narrow) AI.

> **Exam tip:** Always include a justification when asked to classify AI: 1 mark for the correct classification, 1 mark for a valid justification linked to the definition.

## Key Concept: Intelligent Agents

**Intelligent Agent** — An autonomous AI system that perceives its environment via sensors and acts on the environment via actuators, to achieve predefined goals.

*Example:* A self-driving car is an intelligent agent that perceives roads via LiDAR/cameras and acts via steering and brakes.

For 9618 Paper 1, you only need to recall the core definition and the role of sensors/actuators, you do not need to memorize the full hierarchy of agent types.

**Worked example:** For a lawn-mowing robot intelligent agent, name one sensor and one actuator it uses, and state their purpose.

1. Recall: Sensors collect information about the environment, actuators perform physical actions to change the environment.
2. Example sensor: Obstacle sensor, to detect lawn furniture or trees that the robot must avoid.
3. Example actuator: Cutting blade motor, to act by cutting grass to the required height.

## Core Characteristics of Intelligent Systems

You may be asked to explain what makes a system 'intelligent' for 3-6 mark questions. The core characteristics that examiners look for are:

- Ability to learn from experience and adapt behaviour over time
- Ability to reason with uncertain or incomplete information
- Ability to make goal-oriented decisions
- Ability to interpret unstructured input (e.g. natural language, images)
- Ability to solve problems without following fixed pre-programmed rules only

**Check your understanding**

Test your understanding of core characteristics:

1. Which of the following is NOT a core characteristic of an intelligent system?

   - Adapting to new input
   - Following only fixed pre-programmed rules
   - Learning from past errors
   - Handling incomplete data

   *Why:* Fixed rule-based systems cannot adapt or learn, so they do not count as intelligent. Intelligent systems go beyond fixed rules to solve problems.

**Worked example:** A fraud detection system improves its accuracy after processing more transaction data. Explain how this demonstrates intelligent behaviour.

1. One core characteristic of intelligent systems is the ability to learn from experience to improve performance.
2. The system uses past transaction data (experience) to adjust its detection models, reducing false positives over time.
3. This adaptation and learning fits the definition of intelligent behaviour, so the system demonstrates AI characteristics.

## Common Real-World AI Applications

Exam questions often ask you to identify why a given system uses AI, rather than traditional non-AI programming. Common applications you should recognize are:

| Application | AI Type |
| --- | --- |
| Voice assistants (Siri/Google Assistant) | Weak AI |
| Medical image cancer detection | Weak AI |
| Spam email filtering | Weak AI |
| Chess/Go playing AI | Weak AI |
| General human-level AI | Hypothetical (not yet realized) |

**Worked example:** Explain why a spam email filter is an AI application.

1. Spam filters learn to identify new types of spam from labelled training data of spam and non-spam emails.
2. They adapt their rules over time instead of relying on static, unchanging rules written by programmers.
3. This ability to learn and adapt means the system uses AI techniques to perform its task.

## Common pitfalls

- **Wrong:** Claiming AI and Machine Learning are the same thing
  - Why it fails: AI is the broader field, Machine Learning is only one subset of AI approaches
  - Correct: Define AI as the overall field of intelligent systems, and ML as a specific subset of AI that learns from data
- **Wrong:** Classifying modern large language models as strong AI
  - Why it fails: Even advanced LLMs are narrow systems designed for text tasks, they do not have general human-like intelligence
  - Correct: Classify all existing commercial AI as weak/narrow AI; strong AI remains purely hypothetical
- **Wrong:** Forgetting to justifying AI classifications in exam answers
  - Why it fails: Most 2+ mark questions require a justification to earn full marks, not just the correct classification
  - Correct: Always link your answer back to the core definition of the AI type to get all available marks
- **Wrong:** Calling all complex computer systems AI
  - Why it fails: Complex tasks like 3D rendering can be done with traditional non-AI programming, requiring no intelligence
  - Correct: Check for learning/adaptation/reasoning: only systems with these traits are classified as AI

## Cheatsheet

| Concept | Key Definition |
| --- | --- |
| Weak (Narrow) AI | Task-specific AI, no general intelligence |
| Strong AI | Hypothetical general AI with human-like ability across all tasks |
| Intelligent Agent | Perceives via sensors, acts via actuators to meet goals |
| Core AI Traits | Learning, adaptation, reasoning, handling uncertainty, goal-seeking |
| AI vs ML | ML is a subset of AI that learns from data |

## What's next

Now that you have mastered AI fundamentals, you are ready to explore more specific AI topics in the 9618 syllabus. These core definitions are the foundation for all other AI questions on Paper 1, so make sure you can recall and apply them to any scenario. Mastery of these basics will make it much easier to understand more advanced AI concepts like machine learning and expert systems, which build on the classifications and characteristics you learned here. Both topics are regularly tested in CIE A-Level Computer Science exams.

- [AI search techniques](https://www.owlsprep.com/study/cie-9618-u13-ai-search-techniques/)

---

From [OwlsPrep](https://www.owlsprep.com) — free study guides for A-Level, IB, AP and IGCSE, written against the official syllabus. Canonical page: https://www.owlsprep.com/study/cie-9618-u13-artificial-intelligence-fundamentals/