An Introduction to Artificial Intelligence: Basics, Applications, and How to Get Started

Series 40, 2019 Issue 14.

Artificial intelligence has become a hot and widely recognized concept in recent years, driven by real technological progress and media hype. This short article introduces three topics in plain language: basic AI concepts, typical application scenarios, and how beginners can enter the AI field.

1. What is AI?

Many people first heard the term AI from news about the AlphaGo human‑machine match. Although AI (Artificial Intelligence) is a buzzword today, the term was coined half a century ago and has long been expected to bring breakthroughs.

AI is a very broad concept encompassing many technologies. A common question is the difference between AI, deep learning, and machine learning. Below is a diagram illustrating their relative scope: AI is the largest concept, containing both machine‑learning‑based and non‑machine‑learning methods. Non‑machine‑learning methods are older but less popular. Machine learning includes deep learning and non‑deep learning; deep learning, often called neural networks, is currently the most popular.

The most notable AI achievements today are driven by deep learning. What exactly is the “black‑box” algorithm behind deep learning?

Deep learning is often referred to as neural networks, depicted as a picture with many nodes connected together, which may suggest a brain‑like algorithm. In reality, the term is a metaphor; current artificial neural networks have billions of connections, far from the complexity of the human brain. The underlying algorithms are essentially probabilistic statistical methods.

Why has deep learning advanced dramatically only in recent years? The three key factors are data, compute power, and algorithms—often described as “time, place, and people.” Massive internet data provides the “time,” powerful GPUs supply the “place,” and algorithmic breakthroughs give the “people.” The combination of abundant data and super‑computing resources was unavailable decades ago, limiting earlier AI success.

2. What can AI do?

In the internet world, AI has been silently serving us for many years, only recently gaining a formal name. Without AI algorithms, everyday services would be far less efficient.

1. Search: Every day we use search engines like Baidu to quickly find information; without algorithms we would have to sift through countless pages.

2. Recommendation: Platforms such as Toutiao rely on personalized recommendation algorithms to keep users engaged for long periods.

3. Voice: Assistants like Siri and XiaoAi enable direct spoken interaction with machines.

4. Image: Apps like Douyin effects and Meitu rely on AI image algorithms for numerous visual effects.

Other areas include computational advertising, machine translation, and many more.

3. How to get started with AI?

Learning AI mainly involves programming skills and basic mathematics. Python is the dominant language for AI; its simple, human‑friendly syntax allows those with programming experience to pick it up in a few hours, and even non‑programmers can learn it within days.

The mathematical foundation required for an AI beginner is modest—basic statistical concepts are sufficient. Deeper AI research, however, demands stronger mathematical background.

Recommended learning resources:

Textbook: “Machine Learning” by Zhou Zhihua (also known as the “Watermelon Book”), which is clear and approachable.

Video course: Andrew Ng’s Machine Learning course (Stanford), which explains concepts from basics to advanced.

Software library: scikit‑learn , a popular algorithm package with many hands‑on examples suitable for beginners.

Datasets for practice: LIBSVM dataset collection , ImageNet, and others.

Practical experience is essential; by continuously practicing, you can become a proficient AI engineer.