14 Types of AI

AI is far more than Chat GPT. Lets take a brief look at each of the 14 types of AI that have been, and will be, created. These are presented in the order of their broad world-changing power.

This page is for those who are interested - if you want to skip straight to the discussions, you can move now to the next page...

Strong...

Self-Aware AI

The ultimate aspiration of artificial intelligence, self-aware AI refers to systems capable of introspection and understanding their own states. Such entities could not only outperform humans in most economically valuable work but also challenge our perceptions of consciousness and identity.

As well as solving problems, self-aware AI could also understand that it is solving them, reflect on its thought process, and perhaps even have feelings or subjective experiences.

However, developing self-aware AI presents massive challenges, not least because we're still not entirely sure what consciousness is.

Strong AI (also called Artificial General Intelligence - AGI)

Strong AI refers to AI systems that can understand, learn, adapt, and implement knowledge in a way that can independently solve any intellectual task that a human being can. They don't just mimic or simulate human intelligence; they're supposed to possess a form of it.

Strong AI is expected to arrive in 2032. (A year ago, it was expected in 2057.) Self-aware AI is an advanced form of strong AI.

Weak...

There's also "weak AI": all the types below are "weak" examples. Weak is also called "Artificial Narrow Intelligence" (ANI), or "specialised" AI. Weak systems are designed and trained for a specific task.

Theory of Mind (ToM) AI

This class of AI system aims to discern human emotions, beliefs, and motivations (including lying and cheating), providing a deeper connection between machines and our complex internal landscapes. With potential to revolutionise industries from healthcare to customer service, these empathetic machines would comprehend and adapt to human emotions, forging a new frontier in human-computer interaction.

In 2023, Open AI’s GPT 4 solved 95% of theory of mind tasks. It’s nearly here.

Generative AI

The power of AI to generate entirely new content, be it images, text, video, human voices, and even music, is becoming increasingly apparent. Generative AI, such as GANs (Generative Adversarial Networks), could significantly impact fields ranging from art to pharmaceutical development, offering boundless creativity with the precision of machine learning.

Generative AI arrived in 2016, but its impact exploded onto the world in late 2022 with the arrival of Chat GPT. 100 million users signed up in the first 8 weeks.

Cognitive Computing AI

Designed to mimic human brain function, cognitive computing AI strives to create a more natural, human-like interaction between man and machine. From IBM's Watson providing diagnosis advice to doctors, to AI platforms predicting market trends, cognitive computing could redefine how we approach problem-solving across disciplines.

IBM's Watson system is often credited with popularising cognitive computing AI when it won the Jeopardy! game show in 2011. Its latest achievement is providing spoken commentaries on Wimbledon tennis matches.

Swarm Intelligence AI

Inspired by the collective behaviour of natural systems such as ant colonies or bird flocks, Swarm Intelligence AI could revolutionise sectors like logistics, disaster management, or even space exploration. By optimising coordination among numerous AI agents, complex tasks become achievable with remarkable efficiency.

It was first introduced in 1993 by Gerardo Beni and Jing Wang in their paper titled "Swarm Intelligence in Cellular Robotic Systems".

Natural Language Processing (NLP) AI

The ability for machines to understand and generate human language has already transformed how we interact with technology. From personal assistants like Siri to machine translation services, NLP AI holds immense potential to reshape global communication, eliminate language barriers, and offer newfound accessibility.

Significant progress was made in the 1980’s in developing practical NLP applications. The concept arose in the 1950s.

Computer Vision AI

This branch of AI, focused on enabling machines to interpret visual information from the real world, could dramatically change industries from security to healthcare. The promise of self-driving cars, automated medical diagnostics, and intelligent surveillance systems are just a few of the transformative possibilities.

Invented in the 1960s, it wasn’t until the 1990s that real progress was made.

Predictive AI

The ability to forecast future outcomes based on historical data has vast implications across sectors. Whether it's predicting customer behaviours in marketing, anticipating stock trends in finance, or even disease outbreaks in healthcare, predictive AI could make our world more efficient and proactive.

Arthur Samuel developed the first machine learning algorithm in 1959, and Geoffrey Hinton made significant contributions to deep learning algorithms in the 1980s and 1990s.

Limited Memory AI

These AI systems, capable of learning from historical data to make future decisions, are already changing our world. From recommendation systems on streaming platforms to real-time traffic predictions, limited memory AI has become a quiet yet integral part of our digital landscape.

Some of the earliest work on limited memory AI can be traced back to the 1980s when researchers began exploring the use of neural networks for pattern recognition tasks.

Expert Systems AI

These AI systems emulate the decision-making ability of a human expert, providing insights or solutions in specific fields. From medical diagnosis tools to systems for identifying potential cyber threats, expert systems AI provides a level of precision and consistency that enhances human expertise.

The concept of expert systems AI was first introduced in the 1960s by researchers at Stanford University

Interpretive AI

With the ability to interpret and make sense of diverse forms of data – from text to images to complex datasets – interpretive AI is paving the way for more intuitive and context-aware systems. This will transform fields like law, where AI can help make sense of complex legal documents.

Geoffrey Hinton, a pioneer in the field of artificial neural networks, proposed a method in 2006 for training deep neural networks, marking the birth of "deep learning".

Robotic Process Automation (RPA) AI

Automating repetitive tasks has huge implications for business efficiency. RPA AI systems, which mimic human actions to carry out mundane tasks, frees human workers for more complex and creative duties, changing the face of the modern workforce.

The concept of RPA has been around for several decades, but the modern form of RPA was first introduced in the early 2000s by Blue Prism, a UK-based software company.

Reactive Machines AI

The most basic form of AI, these systems react to inputs without using past experiences to inform their actions. From simple chatbots to automated vacuum cleaners to chess-playing computers like Deep Blue, reactive machines AI was the first step into a world where machines can perform tasks independently, laying the groundwork for more advanced AI to come.

The term was first coined by Rodney Brooks in 1986. Brooks developed a robot called Genghis that could navigate its environment without relying on pre-programmed maps or models.