Artificial Intelligence and Machine Learning

Introduction

In the modern era, the terms AI, Machine Learning, GPT and such have been used freely, without many understanding what they actually mean. This post aims to give a brief introduction to AI/ML and how to use it.

Artificial Intelligence

Artificial Intelligence is a broad domain in Computer Science and Engineering where machines and algorithms are fine-tuned to mimic the cognitive process of human thinking. It has reasoning, logical and critical thinking abilities, borne from technologies that allow it to exist as a system on its own; it is currently in research to see if an AI Model can surpass human intelligence and reasoning in major scenarios of the modern world. Popular examples are LLMs, autonomous flight navigation systems, etc.

Machine Learning

Machine Learning is a subset of Artificial Intelligence, and is not as capable as most AI Systems. They are trained on pre-recorded data and are tuned to accomplish a specific task. They do not possess complex reasoning and critical thinking, but excel in providing the output in the task that they are defined for, and improve from experience. Common examples can be a weather prediction system, a driver drowsiness detection system, etc.

What is the relation between them?

The difference between Machine Learning and Artificial Intelligence is noticeable now. As a matter of fact, Machine Learning is a subset of Artificial Intelligence.

One way to think about it is that Artificial Intelligence is the broader sense of machines grasping the concept of independent thinking, making decisions and embracing cognitive behaviour.

Machine Learning, on the other hand, is the application of Artificial Intelligence, where machines extract meaningful data from given data, and learn from it on their own.

Consider vehicles as an analogy. Vehicles are machines that transport something/someone from Point A to Point B. Transportation may be slow or fast, on air, land or through water, it may be on wheels, rails, wings, etc. The general idea is that vehicles are used to transport people/items. Now a sedan is a car that is streamlined, has 4 wheels and travels on road. So while the sedan can transport people, it is limited to just land. However, it does a great job at transporting people on land. In this case, vehicles can represent AI, and the sedan can represent a machine learning model.

What is the difference between them?

By now, it must be clear how different Artificial Intelligence and Machine Learning are.

Artificial Intelligence aims to promote cognitive thinking and decision making in a machine. This encompasses many attributes, skills and factors that also influence a human being’s decision making. In a vague sense, you may consider Artificial Intelligence to be a Jack of all Trades. An AI Model may have various skills over different domains, but they may not necessarily be perfect or masterful in them.

Machine Learning aims to train a machine to accomplish a given task based on past learnt data. While the model may excel and gain experience in accomplishing that task over and over again, it has no other skill to show for itself. In a vague sense, you may consider Machine Learning to be a Master of One Trade.

Let it be noted that while the above statements are true in terms of what we see machines do, they do not truly feel, or analyse or have skills. They are just massive mathematical models that output statements based on probability functions.

Taking this understanding further, it can be understood how Machine Learning is a subset of Artificial Intelligence. The difference between a Jack of all Trades and a Master of One Trade is that the “universe” in which each aim to understand is different. The “universe” of an Artificial Intelligence Model is cognitive thinking on par/subpar to human cognitive thinking. The “universe” of a Machine Learning Model is that one problem that it is designed to solve.

In both cases, the machines are learning from experience, however, what they experience is fundamentally different. We can say, in a rudimentary fashion, that the difference between Artificial Intelligence and Machine Learning is the input data that we give to train the machines.

AI/ML in the Modern World

AI/ML has taken the modern market and ecosystem of today’s world like a storm. Everyone is buzzing around with AI Models, be it commerce, healthcare, telecommunications or even recreational facilities.

This has also changed what the meaning of “intelligent” means. 5–6 years ago, an intelligent system was one which had basic embedded logic and sensor interfacing, allowing for ease of use among consumers. Nowadays, an intelligent system involves an underlying AI/ML Model computing and giving outputs for problems that they are solving.

With the rapid advancements in AI/ML Algorithms, the term “intelligent system” may not be valid, as every system will have some AI/ML Algorithm supporting it. It will be a matter of which model is better than the other. In such a rapid environment, learning how to adapt to it is key to every consumer and producer.

How we can work along with AI/ML

Many critics comment on AI stealing jobs and replacing human workforce, and they call for a reduction on investments in the AI Market. The statistics also support this. As AI Models get progressively better, more and more layoffs have been reported in most big tech companies.

However, it is not just AI/ML Algorithms being better than human thinking. To even get the desired output, a prompt must be given to the AI Model. [Agentic Systems exist, but that’s a story for later]. A human is required to enter that prompt to the AI Model to get the desired output. So now the demand is to be that human who can properly prompt AI Models to get the output.

This is the premise of how Prompt Engineering began. It is basically (and vaguely) how can one prompt an AI Model to get the desired output without wasting time and energy. One way to start with Prompt Engineering is to consider AI Models to be your co-workers, not just tools to use. Delegate tasks and make the final goal, along with the route to be taken, clear and concise. Many courses are available on prompt engineering, for example, Anthropic has a course on AI Fluency, which is a good place to start out if you are a beginner.

Conclusion

While critics still work against AI/ML, it must also be said that the same criticism were against plastic, calculators, and even at one point in time, reading books was considered a bad practice. While their concerns are valid, it should only motivate people to adapt to the situation, rather than to remove the situation.

Thank you for reading till the end!