facts to figures - crack the codes, key AI terminologies

Artificial Intelligence - AI terminologies

(Foreword - I love setting the context before I publish a piece here on LinkedIn for my readers, and hence this introductory paragraph is to just set the reader's expectations. This curated article is not an ordinary one and has not just come from GPTs or other language models etc. This article has dived into various research reports, learning channels along with GPTs to get to this stage to further help AI enthusiasts and our readers to have foundational AI terminologies accessible anywhere, anytime)

Let's start. Guess what? What is AI? what are the key tech stacks (I mean the rocket science) behind it? As AI has been the talk of the town and would remain the same in the coming few years, what if one could start defining AI with zero stumble. I felt why not just dig in a bit and curate a summary of AI fundamentals and key terminologies, which could probably benefit anyone, no matter where they are in their AI journey - learning, adoption or development.

Well, I won't say future is here with AI but the future is not that far too ;) Okay, let me define AI in 6 words:

~ DATA, CONVERTED into HUMAN LANGUAGE with COMPUTE ALGORITHMS

Think of AI as a really smart and intelligent device, which loves to think, learn and speak just like we human.

Technology behind AI - is it really a rocket science?

Obviously no, remember the mainframe times, or the black and white cobol screens, from analogy point of view, it might have felt initially a bit like an alien falling from the sky but it was not, right? While AI can tackle complex problems and deliver substantial business value, its underlying principles are not rocket science, in simple terms, AI offers outcomes that derives from deep learning and predictive modeling on an ocean of data to finally mimic human alike skills!

What exactly is rocket science in this context? ~ the endless possibilities with AI. The scale and complexity of the challenges AI can address.

A comprehensive understanding of an industry's fundamental challenges, technological landscape, and intricate systems is essential for achieving groundbreaking AI breakthroughs. Now, let's talk a bit around how AI actually works? What drives AI?

• Data: Did you know, data is no longer just a valuable resource; it's the new fuel driving our entire world. From images and text to numbers and sounds, data is powering everything around us.

• Algorithms: These are like sets of instructions that tell the AI how to process the data.

• Learning: AI can learn from its mistakes and improve over time. This is called machine learning.

  
  

...Now, let's jot down some of the basic yet important AI terminologies with EXAMPLES...

Agents in AI is a system that can perceive its environment and take actions to achieve specific goals - e.g. self-driving car navigating a road or a chatbot answering customer questions.

Assisted Intelligence refers to AI systems that work alongside humans to enhance their capabilities - e.g. a virtual assistant helping a doctor diagnose patients or a predictive analytics tool assisting a financial analyst in making investment decisions.

Backward chaining is a reasoning technique that starts with a goal and works backward to find the conditions that must be true for that goal to be achieved. For example, to prove "if it's raining, the streets will be wet," backward chaining would start with the goal "streets are wet" and work backward to determine if "it's raining" is true.

Augmented intelligence is similar as assisted intelligence with a slight difference, augmented intelligence is a collaboration between humans and AI systems to enhance problem-solving and decision-making. For example, a radiologist using AI to analyse X-rays for early detection of cancer. Unlike, assisted intelligence, augmented intelligence, goes a step further where AI actively collaborates with humans to enhance decision-making and problem-solving.  For example, an AI assistant that suggests email replies is an example of assisted intelligence, while an AI system that analyzes medical images and provides insights to a doctor is augmenting their intelligence and hence an example of augmented intelligence.

Machine Learning is a subset of AI that involves training algorithms on data to learn patterns and make predictions, such as recognizing images, predicting stock prices, or recommending products.

Deep Learning is a subset of machine learning that uses neural networks with multiple layers to learn complex patterns from data, such as recognizing images, understanding natural language, and playing games.

Neural Networks are a specific type of machine learning algorithms and computational model, inspired by the human brain, used to recognize patterns and make predictions in tasks like image recognition, natural language processing, and self-driving cars.

Generative AI is a type of artificial intelligence that can generate new content, such as text, images, audio, or code. It's trained on massive datasets and uses complex algorithms to learn patterns and structures within that data. Once trained, it can produce new content that is similar to the training data but not identical.

Hallucination in AI refers to the phenomenon where an AI model generates content that is factually incorrect or misleading, often based on incorrect assumptions or misunderstandings of the data it was trained on.

LLMs (Large Language Models) are a specific type of machine learning model, trained on large amounts of texts, designed to understand and generate human language and text base content, while machine learning encompasses a broader range of algorithms for tasks like image recognition, predictive analytics, and more. For example, GPT-3 is an LLM, while a model used to predict stock prices would be a machine learning model.

NLP (Natural Language Processing) is a field of AI that focuses on the interaction between computers and human language, enabling tasks like machine translation, sentiment analysis, and chatbots. NLP is a subset of ML which focuses specifically on understanding and generating human language, while machine learning encompasses a broader range of algorithms for tasks like image recognition, predictive analytics and many more. For example, a chatbot or a sentiment analytics tool that can understand and respond to customer inquiries uses NLP, while a model used to predict stock prices would be a ML model.

Okay, let's look at some realistic numbers now, where AI is heading and what is the potential opportunity or future with it:

As per McKinsey's April 2024 forecast:

~ 60 countries currently have national AI strategies ~ 2300 , the year AI capabilities will rival humans ~ $4.4 trillion annually gen AI could add to the global economy

Also, as per Statista,

The market size in the Artificial Intelligence market is projected to reach US$184.00 bn in 2024 with a CAGR of 28.46%, resulting in a market volume of US$826.70 bn by 2030. In global comparison, the largest market size will be in the United States (US$50.16bn in 2024).

Now, can you imagine what's in for businesses leveraging AI or building with AI or even scaling with AI.

Before we wrap - AI Evolution

Now, before I stop adding more to this article, let's quickly understand how AI has evolved - how AI has been in there with us for long and used by companies but got the real kick in this era with GenAI, new automation use cases and advanced predictive modeling tools. From its early experimental stages to becoming an integral part of our modern world, from the initial development of expert systems to the recent breakthroughs in deep learning, AI has demonstrated remarkable progress. Let's dive into it a bit and see how far we have come:

Early Foundations

• 1950s: Alan Turing's "Computing Machinery and Intelligence" introduces the concept of the Turing Test, a measure of a machine's ability to exhibit human-like intelligence.

• 1956: The Dartmouth Summer Research Project on Artificial Intelligence marks the official birth of AI as a field of study.

  
  

Early Triumphs and Challenges

• 1960s-1970s: Development of expert systems, problem-solving programs designed for specific domains like medical diagnosis and financial analysis.

• 1970s: The "AI Winter," a period of decreased funding and interest due to limitations of early AI systems.

  
  

Resurgence and Advancements

• 1980s: Resurgence of AI with advancements in hardware and software, leading to the development of neural networks and machine learning algorithms.

• 1990s: Commercialization of AI applications, including expert systems and natural language processing.

• 2000s: Advancements in machine learning, particularly support vector machines and decision trees.

• 2010s: Breakthroughs in deep learning, with convolutional neural networks (CNNs) and recurrent neural networks (RNNs) achieving significant success in various tasks.

  
  

Modern AI

• 2010s-present: Continued advancements in deep learning, leading to breakthroughs in image recognition, natural language processing, and speech recognition.

• 2010s-present: Development of AI-powered systems for self-driving cars, medical diagnosis, and personalized recommendations.

• 2020s: Increasing focus on ethical considerations in AI, including bias, privacy, and job displacement.

  
  

As AI continues to advance, it is crucial to address ethical concerns and ensure its responsible development and deployment for the benefit of society.

That's pretty much for now. We've got another game-changing AI topic already brewing. So stay tuned and we will be back soon with the next read therapy. Until then, stay healthy and happy.

For anything AI, we are just an email away, write us at consulting@stunm.in or book a consulting right here.