Image by Editor | ChatGPT

# Introduction

 
Everyone talks about artificial intelligence (AI) in recent times, and chances are, depending on whom you ask, they may explain to you what AI is in a different way: often using technical, almost alien-like terms, and sometimes in simple terms. This article is intended to take the latter direction, and explain in simple words what AI is, its capabilities and limits, and some key areas or building blocks that are part of AI nowadays.

# What is AI? Capabilities and Limits

 
AI is a branch of computer science that focuses on the development of computer systems capable of mimicking human or animal intelligence skills to solve a particular problem or task. The intelligent traits they can emulate, depending on the specific AI system, include reasoning, predicting future or unknown facts, visual perception, communicating with humans or other AI systems, and interacting with the environment, to name a few.

One salient capability in most modern AI systems, which permeates virtually all the skills mentioned above, is the ability to learn from data. Just like a small child who learns by observing and being exposed to examples of anything (words, images, actions, etc.), AI systems learn from lots of data examples to perform complex tasks like predictions, inference, communication in human language, and much more.

Some might be surprised to hear that AI is not a 21st-century invention: in fact, the field was born in the middle of the last century. However, it is thanks to advances in learning from data — and thanks to developments in computation and data availability — that AI has experienced its biggest progress over the last decade and a half, with generative and language models like ChatGPT being clear examples of the latest, high-impact achievements.

But AI also has its limitations: it does not truly “understand” the world like humans do (everything is understood and dealt with mathematically, so to speak!), can make mistakes if it has learned from poor-quality data, and usually struggles with tasks that require common sense or human sensibilities. The vast data and computing power needed also constitute high costs.

# Areas of AI

 
Let’s outline the main areas of AI and some key points to gain a better understanding of them.

Areas of artificial intelligence and common symbiotic relationships between them | Image by Author

• In the early years of AI, knowledge representation and reasoning were active subjects of research, focuses on building software programs that could reason and draw conclusions upon real-world facts and knowledge represented in a machine-understandable fashion. Much of the logic between autonomous agents (sometimes implemented in robotic systems capable of interacting with the real world) relies on these techniques, and the recent rise of agentic AI solutions is reviving interest in this subdomain.
• Machine learning is by far the most widespread area of AI at present: almost every modern AI system uses approaches to learn from data to solve a task like classifying images, predicting prices or future sales, recommending products to customers, and more.
• Deep learning is like an “evolved form” of machine learning: larger and more complex systems that learn from immense datasets to address more challenging tasks. Interestingly, computer vision and natural language processing applications — like visually recognizing objects and having a conversation with a human — have recently become predominantly solved using deep learning systems, hence their close relationship.
• So, where do the latest trends of generative AI and language models (LLMs) sit? These systems are massive deep learning architectures taken to the next level.

That was a concise, gentle introduction to what AI is. Hope you enjoyed it!
 
 

Iván Palomares Carrascosa is a leader, writer, speaker, and adviser in AI, machine learning, deep learning & LLMs. He trains and guides others in harnessing AI in the real world.