BACKGROUND IMAGE: iSTOCK/GETTY IMAGES
When we talk about AI today, we're not talking about one technology. In fact, AI tools are made up of several discrete components that combine to create the applications we typically think of as AI.
So what are the primary AI components? It starts with a foundation of specialized hardware and software for writing and training machine learning algorithms.
Deep learning, a specific branch of machine learning, takes an intuitive approach to teaching algorithms to process data that is similar in some ways to how a human learns. The approach can deliver impressive results, but the downside is that it takes massive amounts of data to train algorithms. That's why data scientists need the processing power of GPUs to train models and keep training times down.
No one programming language is synonymous with AI, but a few have set themselves apart. Python is quickly becoming a popular, general-purpose language for writing analytics applications, so it's no wonder it's becoming common in AI applications. Java and C are also widely used because they offer deep, low-level control, though this can make them more complicated than Python. Google's open source TensorFlow is also gaining some traction as a component of AI.
Today, we're seeing AI applications take a wide variety of forms. They are doing everything from translating websites to serving as personal assistants.
The general theme of most AI services, though, is an ability to understand human language and interpret its often vague meaning. True AI applications also use machine learning to take new information into account and use it to sharpen its performance over time.
As AI hype has accelerated, vendors and tech evangelists are calling all kinds of things AI. Often what they refer to as AI is simply one component of AI, such as machine learning.
While the technologies described here are far from an exhaustive list of AI components, the list does suggest that AI is more complicated than any single technology. AI is a collection of tools and techniques.
Today's AI holds promise, but we've only seen the tip of the iceberg
Don't let biased data ruin your AI projects
Businesses need to get ready to roll out AI applications
AI tools find a home in the modern data center