An overview of the different types of current artificial intelligences. Uses and applications.

Artificial intelligence (AI) is the ability of machines or computer systems to perform tasks that normally require human intelligence, such as pattern recognition, learning, decision-making or natural language processing. There are different types of AI depending on their degree of autonomy, complexity and scope of application. In this article we will give a brief introduction to the main types of AI that exist today and some examples of their uses and applications.

Types of AI according to the degree of autonomy

Depending on the degree of autonomy, we can classify AI into three types: assisted, augmented and autonomous.

• Assisted AI is AI that relies entirely on human intervention to function and perform its tasks. It is the simplest and most widespread today. Examples include virtual assistants such as Siri or Cortana, recommendation systems such as Netflix or Amazon, or facial recognition systems such as Face ID or Facebook.

• Augmented AI is AI that can operate independently, but can also interact with and learn from humans. It is more complex than assisted AI and has greater potential to enhance human capabilities. Examples include autonomous driving systems such as Tesla or Waymo, medical diagnostic systems such as IBM Watson or Google DeepMind, or machine translation systems such as Google Translate or Microsoft Translator.

• Autonomous AI is AI that can operate without any human supervision or intervention, and that can make its own decisions and take its own actions. It is the most advanced and the most difficult to achieve, as it involves endowing machines with consciousness, creativity and ethics. Examples include humanoid robots such as Sophia or Atlas, intelligent agents such as AlphaGo or OpenAI, or content generation systems such as GPT-3 or DALL-E.

Types of AI according to complexity

Depending on complexity, we can classify AI into two types: weak and strong.

• Weak AI is AI that is designed to perform a specific and limited task, and cannot generalise or transfer its knowledge to other tasks. It is the most common and the most accessible today. All of the above examples belong to this type of AI.

• Strong AI is AI that is designed to perform any task that a human can do, and that can generalise and transfer its knowledge to other tasks. It is the most ambitious and the furthest into the future. There is no real example of this type of AI, although some theorists speculate on the possibility of creating an artificial superintelligence that surpasses the human in every respect.

Types of AI by application domain

Depending on the scope of application, we can classify AI into four types: reactive, memory-based, theory-based and self-aware.

• Reactive AI is AI that only responds to stimuli in the environment, without memory or prior knowledge. It is the most basic and the oldest. An example is the Deep Blue program that defeated world chess champion Garry Kasparov in 1997.

• Memory-based AI is AI that stores data and information from the past, and uses it to improve its performance in the present. It is more sophisticated than reactive AI and is based on machine learning. An example is the AlphaGo programme that defeated world go champion Lee Sedol in 2016.

• Theory-based AI is AI that builds models and representations of the world, and uses them to reason and plan its actions. It is more advanced than memory-based AI and relies on automatic reasoning. An example is the Cyc programme that attempts to create a common knowledge base for all areas of human knowledge.

• Self-aware AI is AI that has a sense of itself, its goals, emotions and values. It is the most complex and the most unknown. There is no real example of this type of AI, although some researchers are working on the development of emotional intelligence and social intelligence for machines.

Conclusion

Artificial intelligence is a very broad and diverse field, encompassing different types, levels and applications. Its evolution and progress depend on multiple factors, such as technology, science, ethics and society. Its potential is enormous, but so are its challenges and risks. It is therefore important to understand the different types of AI that exist today, their uses and their implications, in order to take advantage of their benefits and avoid their dangers.