Artificial Intelligence (AI) is a field of computer science that deals with the development of intelligent machines that can perform tasks that typically require human intelligence, such as visual perception, speech recognition, decision-making, and language translation. AI systems are designed to learn from data, identify patterns, and make predictions or decisions based on that data. AI has the potential to transform many aspects of our lives, from healthcare and transportation to education and entertainment. It has the potential to solve some of the world’s most pressing problems, such as climate change, poverty, and disease. However, it also raises ethical and philosophical questions, such as the impact of AI on jobs, privacy, and autonomy.

When it comes to defining AI, one can broadly define it in three ways:

1. “Cool things that computers can’t do” (1) is a very vague statement and cannot be used to define AI. Terms like “Cool” are subject to interpretation. Also, computers (in traditional sense) do not have appendages (in other words typical sensors) so for example, a computer making me a coffee when I am tired is “Cool” but still cannot be attributed to AI. All the various functions that seem “Cool” can be broken down into a set of rules based on data collected and interpreted by the sensor.

2. “Machines imitating intelligent human behaviour” (2) is a very generic but useful definition of AI. Though the term “intelligent” is left to interpretation and differs based on situational awareness, experience, knowledge, approach to problem statement, etc. As with humans, there can be multiple levels of AI, but still is a good definition of AI.

3. “Autonomous & Adaptive Systems” (3) can at first glance look like AI, but these are applications of AI. A sufficiently intelligent system can be adaptive based on defined criteria and can generate autonomous responses. So (3) is an application of AI and cannot be termed AI itself. (3) is also based on model training and model inference. So a model based on faulty/wrong data will generate an inference that is invalid (from a general perspective) but valid based on the data training. In essence, I still see this as a mesh/inter-working of numerous atomic functions in a very complex chain of execution relying on input and output data points.

I will define AI as a “consciousness” or “intelligence” that:

• recognizes, organizes, validates and interprets data from any and all sources
• can learn from data and improve its performance based on the data
  — can adapt to new situations and make decisions based on the data
  — can communicate with other intelligent entities
  — can create new intelligent entities
  — can detect right from wrong and make rational decisions based on collective understanding of moral principles
  — Understands the fight or flight response mechanisms (in other words, can compete and work for its own survival)
  — Uses “experience”, keeps the context alive and realizes a purpose for survival

So the above talks about a lot of characteristics of AI in general. A “limited” can maybe satisfy some criteria but a truly wholesome general purpose AAI (Advanced Artificial Intelligence) maybe can satisfy all.

AI can be classified by:

Technology:
–> Machine Learning which includes
–> Supervised Learning
–> Unsupervised Learning
–> Semi-Supervised Learning
–> Reinforcement Learning
–> Deep Learning
–> Natural Language Processing (NLP)
–> Large Language Models (LLMs)
–> Computer Vision
–> Speech Recognition
–> Robotics
–> Expert Systems

Capability:
–> Weak AI (Narrow AI)
–> Strong AI (General AI)
–> Super AI (Superintelligent AI)

Functionality:
–> Reactive Machines
–> Limited Memory
–> Theory of Mind
–> Self-Aware AI

Purpose:
–> Generative AI
–> Generative Adversarial Networks (GANs)
–> Variational Autoencoders (VAEs)
–> Diffusion Models
–> Multimodal AI
–> Recurrent Neural Netowkrs (RNNs)
–> Flow Models
–> Neural Radiance Fields (NeRFs)
–> Reactive AI
–> Predictive AI
–> Assistive AI
–> Conversational AI

Details on the above can be found on AI Types and AI Capabilities on Wikipedia. A quick summary is also available here.

The future is going to expand on these definitions, bringing in new dimensions to the AI space. How would you define AI?

Do write your comments and expand on the thought process.