In an age where technology evolves at breakneck speed, in 2024, Artificial Intelligence (AI) stands at the forefront, leading the charge. Previously confined to filling pages of science fiction, it now splashes across news headlines almost daily, but what does it all really mean? In part one, we covered essential AI terms from A to L. Now, we’re back to look at the key concepts, from M to Z…

 

Machine Vision

Machine vision is AI’s answer to human sight. It enables computers to make sense of visual information from the world. Similar to how we use our eyes to identify objects and assess situations, machine vision allows machines to process, analyse, and understand images and videos for a variety of applications, from quality control in manufacturing to facial recognition systems and the creation of AI avatars.

 

Neural Networks

At the heart of many AI systems, especially deep learning, are neural networks. These are algorithms modelled on the human brain’s structure and function, capable of learning from vast amounts of data. They are networks of interconnected nodes (like neurons, aka brain cells) that can provide ‘super-human’ pattern recognition across huge quantities of data, and use this to make decisions.

 

Quantum Computing

Quantum computing is a growing field that promises to revolutionise AI by using the principles of quantum mechanics (the study of how subatomic particles behave and interact to process information) to process information at speeds that are almost incomprehensible by today’s standards. While still very early technology, quantum computing could one day enhance AI’s learning and processing capabilities exponentially.

 

Robotics

Robotics, often intertwined with AI, is the field concerned with creating and programming robots—autonomous machines that can perform tasks in the real world. When powered by AI, robots can learn and adapt to their environment, making them incredibly useful in industries ranging from manufacturing to healthcare and beyond.

 

Supervised Learning

This type of machine learning (see part one) is akin to a student learning under the guidance of a teacher. Supervised learning algorithms are ‘trained’ on labelled data – imagine thousands of pictures of cats and dogs, each one labelled ‘Cat’ or ‘Dog’ being shown to the computer. This means they learn to predict the correct output from this input data, allowing them to make predictions or recognise patterns when given new, similar datasets.

 

Unsupervised Learning

In contrast, unsupervised learning involves algorithms that learn from data without explicit instructions on what to do with it. They are adept at identifying complex patterns and structures within datasets where the ‘right answers’ are not provided. This is how many of the chatbot systems like ChatGPT were trained – taking in vast quantities of data from the web (that would be almost impossible to label manually).

 

Virtual Reality (VR)

VR is a simulated experience that can be similar to the real world… or entirely different. VR generally utilises a headset that users wear to immerse themselves in the simulated world. AI enhances VR by making the environments more interactive and responsive to the user’s actions, providing an immersive experience in gaming, training simulations, and even therapeutic settings.

 

XAI (Explainable AI)

Explainable AI is the movement towards more transparent AI, where the decisions and processes of AI systems can be understood by humans. Currently, decisions and responses made by AI systems are often not fully explainable by the developers (for example, the scientists behind Chat GPT cannot say precisely why particular words and phrases are used in a response). Some see Explainable AI as crucial for maintaining trust and accountability as AI systems become more prevalent in critical decision-making roles.

 

eXistential Threat

Amidst the rapid advancement of AI, a term that often surfaces in discussions is “existential threat.” This refers to concerns that AI could one day pose significant risks to human existence or drastically alter life as we know it. Some AI scientists fear that as AI becomes more sophisticated, it could outpace human intelligence, leading to a future world where machines operate beyond our control. The worry stems from the potential of increasingly powerful AI systems to autonomously make decisions that could have significant impacts on global security, economics, and privacy. Addressing these risks involves governments, AI companies, academics, and others collaborating to develop safety procedures, testing programmes and new regulations to ensure AI development aligns with human values and well-being.

 

Zero-shot Learning

A leap in machine learning, zero-shot learning refers to the ability of an AI to recognise objects or understand tasks it has not seen during its training. It’s about applying knowledge learned from one context to a completely new context, much like a musician who can play a song they’ve never seen or heard before because they have learned to read music from other songs.

 

Understanding these terms is just the start of the journey into the fast-moving world of AI. From the way machines ‘see’ and ‘think’ to how they learn and make decisions, the impact of AI is far-reaching. As we wrap up our glossary series, remember this is just the beginning. Many resources are available online if you want to learn more about AI – simply Googling words and phrases can be a good start. The field of AI is evolving at a rapid speed, promising even more exciting developments on the horizon, so stay curious about the possibilities.