Series 1: AI Basics — Chapter 1: Fundamentally Understanding AI
The first of seven chapters aimed at simplifying AI for people without a technical background.
This is the 1st of seven articles that are dedicated to helping non-technical readers understand Artificial Intelligence (AI) and what it could mean for them.
As a non-technical person myself, I plan to share what I’m learning along the way and will do my best to make these seemingly complex topics simple.
With a lot of information out there, I’ve structured my writing into multiple Series (categories of topics) and Chapters (concepts within each topic). This is the first series on ‘AI Basics’ and contains seven chapters.
Now, let’s get into it.
Chapter 1: Fundamentally Understanding AI [this article]
Chapter 2: Differences between AI, ML, DL, GenAI & LLMs
Chapter 3: A Brief History of Artificial Intelligence
Chapter 1: Fundamentally Understanding AI
This article covers:
1. What is Traditional Programming
2. What really is Artificial Intelligence
3. Advantages and disadvantages of both these disciplines
4. Conclusion
So, how do you, a non-technical person, understand what AI really is? Well, the best way to understand AI is to know how it differs from traditional programming.
But first, what is Traditional Programming? After all, this article is written for non-technical folks!
Traditional Programming is done by defining a set of rules or logic for a desired outcome. For example, to perform the function A + B = Y, a programmer must define the rules to add the two numbers A and B as ‘inputs’ to provide an ‘output’ called Y. Based on these rules, the computer will perform an addition of the two numbers and provide the resulting output.
This is a manual process, and so whenever there is a change to the desired outcome, the rules have to be rewritten. For example, if the desire is to now do a multiplication, the programmer has to make the necessary changes to the code.
Things to know about traditional programming:
- It requires explicit rules to be manually written by the programmer.
- It’s a deterministic approach to programming like a recipe with step-by-step instructions for preparing a dish.
- It’s for clearly defined problems with a limited number of outcomes.
- With traditional programming, it becomes nearly impossible to write rules for every single scenario when tasks are complex or need human-like perception such as image recognition. These scenarios are firmly in the AI wheelhouse.
What really is Artificial Intelligence (AI)?
Fundamentally speaking, AI is a high-level discipline advancing the simulation of human intelligence in machines through a set of approaches aimed at performing tasks such as visual perception, speech recognition, decision-making, and language translation — skills we humans possess.
Instead of writing rules or logic as we read above in traditional programming, AI requires three components to run effectively — Data, Algorithms, and Computing power, aka Compute.
- Data: AI requires three kinds of data to run efficiently — (i) Training data that helps the AI model learn, (ii) Validation data that tune the model, and (iii) Test data to assess the model’s performance.
- Algorithms: These are the sets of rules to process data and make decisions. (i) Machine Learning algorithms make decisions without explicit programming as they learn from patterns and (ii) Reinforcement Learning algorithms perform functions and receive punishment and awards based on their correctness. More on this in the future.
- Compute: Computing is done using Graphics Processing Units (GPUs) which streamline processes taking in the data and running algorithms.
Advantages & disadvantages of Traditional Programming and AI
It’s important to look at three aspects when comparing these two programming methods:
- Stability and Scalability:
Traditional programming is stable as it performs operations the same way every single time. However, the stability of rule-based programs comes at the expense of scalability since traditional programs can only learn through explicit programming. The bigger the scale of the operations, the more code has to be written.
On the other hand, AI programs offer more scalability than traditional programs but with less stability. The automation and continuous learning features of AI programs allow developers to scale processes quickly and with relative ease. However, the nature of AI programs to improvise may not always provide a consistent and appropriate response.
2. Control and Transparency:
Traditional programs offer developers full control over the logic and behavior of software. If a program doesn’t behave as expected — for example, it has a bug — developers can trace back through the code base to identify and correct the issue.
AI systems make it harder to pinpoint where the issue lies. Complex AI systems like deep neural networks work like black boxes, where developers would only know the input and the output but not how the model arrived from one to the other. This lack of transparency can be problematic in industries that prioritize processes and decision-making like healthcare and finance.
3. Learning and Data Handling:
Traditional programs are rigid as they rely on structured data to execute programs. To ‘teach’ a program new information, a programmer must manually add new data or adjust old processes. Traditional programs also may not accommodate unforeseen scenarios without explicit programming for those cases.
On the other hand, as we saw above, AI systems learn from vast amounts of data so they’re better suited for processing unstructured data like images, videos, and natural language text. AI systems can also learn continually from new data and experiences allowing them to improve performance over time.
Conclusion:
It’s not like traditional programming is going away anytime soon. AI systems are also not without their complications, which we will discuss in future posts.
However, the use of AI presents us with an opportunity to keep pace with an increasingly complex and dynamic world by meeting it with sophisticated technologies that can handle real-world complexity. After all, the term AI was coined in 1956 and here we are 68 years later — do you think it’s time for humanity to make an exponential move forward?
Next up, Chapter 2: Differences between AI, ML, DL, GenAI & LLMs
