Modern approaches to artificial intelligence

Artificial Intelligence (AI) has had an extraordinary journey. From the early days of the Turing test and symbolic reasoning – which rose to prominence in the 1950s – to the 21st century’s cutting-edge advancements in deep learning and neural networks, the field of AI has witnessed remarkable progress. 

Today, AI is embedded in our lives in a myriad of ways – across both business processes and people’s daily routines – and experts agree that emerging areas of artificial intelligence, such as generative AI, will further cement its role in society.

But to better understand the future, it can help to first look back. Exploring and understanding the historical approaches to AI highlight how they paved the way for the modern approaches that are shaping the landscape of AI now. And by understanding the foundations and evolution of AI, we can better appreciate its transformative potential for the years ahead.

Historical approaches to artificial intelligence

The quest for creating intelligent machines dates back to the early days of computer science. In the mid-20th century, the focus was on developing algorithms and formal logic to simulate human intelligence.

This era saw the emergence of symbolic AI, which aimed to represent knowledge and solve problems using rules and logical operations. First-order logic and knowledge representation played pivotal roles in this approach, allowing AI systems to reason and make decisions based on logical inference, and pseudocode and early programming languages like LISP were used to implement AI algorithms.

As AI research progressed, the limitations of symbolic AI became apparent, especially in dealing with uncertain knowledge and complex real-world scenarios. To tackle these challenges, probabilistic reasoning and Bayesian approaches – based on the mathematical theorem of Thomas Bayes – gained prominence. Other advances included probabilistic graphical models, such as hidden Markov models, which enable AI systems to handle uncertain data and make informed decisions under probabilistic constraints.

Modern developments in artificial intelligence

Machine learning and neural networks

One of the most significant breakthroughs in AI came with the advent of machine learning and neural networks.

Machine learning revolutionised the practice of artificial intelligence by enabling systems to learn from data and improve their performance over time.

Neural networks, meanwhile, are inspired by the human brain, making connections between computer nodes the way the human brain makes connections between neurons.

These technologies have formed the backbone of deep learning, which is a sub-area of machine learning and neural networks, one that aims to train AI models with multiple layers and enable machines to learn in a way similar to human beings. In this way, AI can identify and extract complex patterns and representations from data.

The paradigm shift created by machine learning and neural networks has allowed AI systems to tackle a wide range of problem-solving tasks, from search algorithms to computer vision and natural language processing.

Computer vision and natural language processing

Computer vision, a subfield of AI, has witnessed tremendous advancements in recent years. By leveraging deep learning techniques and neural networks, computer vision systems can now accurately interpret and understand visual data, enabling applications such as object detection, image recognition, and autonomous driving.

Natural language processing, meanwhile, teaches AI systems to understand semantics and generate human language. From chatbots to language translation, natural language processing has opened up new frontiers in human-computer interaction.

Reinforcement learning and decision making

Reinforcement learning, a branch of machine learning, takes inspiration from behavioural psychology to train AI agents through interactions with their environments. By rewarding desirable actions and penalising undesirable ones, reinforcement learning algorithms can learn optimal decision-making strategies.

This approach has proven to be instrumental in developing AI systems that excel in complex decision-making tasks, such as playing games like chess and Go, or controlling autonomous vehicles.

Multi-agent systems and AI collaboration

The notion of intelligent agents working together in a collaborative manner has been gaining traction in the field of AI. Multi-agent systems aim to develop algorithms and frameworks that enable multiple AI agents to interact, communicate, and coordinate their actions in order to achieve common goals. This approach has broad applications, ranging from distributed robotics to supply chain optimisation, where multiple agents must cooperate to solve complex problems efficiently.

Future directions and challenges

As AI continues to evolve, new frontiers and challenges emerge. Probabilistic programming, for instance, aims to combine the power of probabilistic reasoning with traditional programming, opening up opportunities for building AI systems that can reason and make decisions in uncertain conditions. Additionally, the field of AI ethics is gaining prominence, addressing concerns related to bias, fairness, and accountability in AI algorithms and systems.

Today, society stands at the forefront of AI research and implementation. By embracing modern approaches to AI, people and organisations can harness its potential to tackle both simple decisions and complex decisions, improve problem-solving capabilities, and advance various fields, from health care to environmental sustainability. 

Further resources

For more information about the history of artificial intelligence, thoughts on where the field is going, and how it’s being used today, there are a number of resources available:

• Open AI Resources, an online directory of open source software and open access data for the AI research community. The site is currently managed by AI Access Foundation, and it includes tools in a variety of languages, including Java and Python.

• Resources for the Field of Artificial Intelligence, an online repository of AI career information managed by the Institute of Electrical and Electronics Engineers (IEEE Computer Society).

• Artificial Intelligence: A Modern Approach, a best-selling and commonly cited AI textbook written by Stuart J. Russell and Peter Norvig, and published by Pearson Education (formerly Prentice Hall). Stuart Russell is a computer science professor at Berkeley, and Peter Norvig is a distinguished education fellow at Stanford’s Human-Centered Artificial Intelligence Institute, and a researcher at Google. There are four editions of the textbook: the 1st edition was published in 1995; the 2nd edition was published in 2003; the 3rd edition was published in 2009; and the most recent version, the 4th edition, was published in 2020.

The text covers topics spanning constraint satisfaction problems and logical agents to the future of AI. For more information about the text, including its US and global editions, ISBNs, and table of contents, visit aima.cs.berkeley.edu.

Develop expertise in modern artificial intelligence

The use of artificial intelligence within the modern business landscape continues to grow at speed as organisations realise the impact it can have on their future success. Specialise in this rapidly evolving field with the MSc Computer Science with Artificial Intelligence at Abertay University. This flexible master’s degree is taught part-time and 100% online, and is suited to mid-level professionals who want to move into computer science, or take the next step in their existing computer science career.

You will develop specialist knowledge in artificial intelligence, including how to use it to enhance processes and how businesses and systems can be improved through AI techniques. One of your key modules – in emergent technologies – will enable you to critically appraise next-generation technologies and their impact on society, and you will consider the role of AI in application deployment.

You will also gain a firm foundation in the field of computer science, with core knowledge in areas such as networking, databases, and web development.