A Beginner’s Guide to Artificial Intelligence

There’s no denying the tumult, horrors and massive loss of life during World War II. But, as is often the case, from this environment of desperation and hardship sprang many technological advances. Alan Turing, helped lead an unprecedented cadre of mathematicians and logicians to success in building a special purpose computer which could break the secret communication codes of their military enemies. Turing emerged from this period inspired to consider the future possibilities of more generalized computing devices and where that could lead society and how those machines would evolve. Turing wrote a paper in 1950 titled “Computing Machinery and Intelligence,” which he opened with the statement: “I propose to consider the question; can machines think?”. It was a ground-breaking paper not least because it contained a test which Turing originally called “The Imitation Game” but which later became more popularly known as “The Turing Test”.

The Turing Test laid out, for the first time, a foundation for how we might consider a machine – a computer – to be “intelligent”. The main premise of the test is that if we cannot tell if we are speaking to a machine or a human – or if we guess and guess wrongly – then, put simply, that machine is probably best considered ‘intelligent’. It’s a simple definition but that simplicity belies the genius of it.

How have things changed since Alan Turing’s time?

Computing technology- and along with it artificial intelligence – has grown exponentially since Turing’s time. Yet we are still only standing on the first rung of a very long ladder.

Picture this; a machine that can organize your portable home library alphabetically, just as you’d like. Or a machine that could prepare a customized daily schedule for everyone at the office. Makes your life/job easier, doesn’t it? These are the products of artificial intelligence. 

But why is it termed “artificial intelligence”? Well, these machines are designed using complex algorithms and mathematical functions that aim to incorporate human-like intelligence, to make decisions and solve problems as we do. However, AI may not be as obvious as in the above examples. In fact, AIs have various applications in almost every area of life. You can find AI in smartphones, cars, wrangling our social media feeds, playing (and beating us) in games, automating banking decisions and anti-fraud or money-laundering activities, widespread private and government surveillance, even performing roles in our vehicles to enhance safety and usability. The real question, however, is what does an AI do at its core?

What makes a product or machine artificially intelligent?

Three basic components make up this answer. We’ll describe them using this illustration. 

Let’s assume we build our robot in a lab and transport it to a park. Regardless of the difference in lighting and landscape between our lab and the park, the robot must perform as expected. This ability to adapt quickly to a new situation is called “generalized learning.” The robot then gets to a crossroad; one, muddy and the other paved. Here, it must determine which path to take based on the circumstances. This portrays the robot’s “decision-making” ability. After a few meters on the paved road, the robot approaches a body of water it cannot swim through. Using a plank added to the robot as input, it can cross to dry land. Here, our robot utilizes its given input to “solve a problem.” 

These three capabilities make the machine artificially intelligent. In other words, AI provides machines or software with the ability to reason, adapt, and provide solutions. Now that we know what AIs are let’s look at the categories they’re classified into.

What are the categories of artificial intelligence?

AI is classified into two broad categories; weak AI and strong AI: 

Weak AI

This is the most common category of AI and is also known as narrow AI, as it focuses solely on one task. For example, Alphago (AI) is a maestro of the game go, but you can’t expect it to be even remotely good at other games like Chess or Ludo. This makes Alphago a weak AI. You’re probably thinking, “Alexa is definitely not a weak AI because she/he can perform different tasks.” However, that’s not the case. For example, when you ask Alexa to play a song by Justin Bieber, it picks up the keywords; “play” and “Justin Bieber” and runs a program with its given input. Alexa cannot provide answers for problems it is not trained to solve. For example, Alexa cannot provide you with traffic status from your home to work unless you have access to an Alexa “skill” for that purpose. A skill in that context just means “a program which allows Alexa to understand what you’re asking and react to it with various pre-programmed actions”. That brings us to our second category

Strong AI

Strong AI doesn’t mean “strong enough to crush your car with a flick of its wrist”. Strong AI is much like the robots and computers that only exist in fiction as of now. Ultron from Avengers is an ideal example of a strong AI. That’s because it’s self-aware and eventually even develops emotions. This makes the AI’s response unpredictable, more like humans. Strong AI has the ability to adapt and grow its intelligence beyond those actions for which it is programmed. In some ways we might liken this to “intuition” or even “evolution”.

Right now AI, as far as we know, all available AI is essentially weak AI – but the time when AI will become truly intelligent is almost within our grasp, decades at most, perhaps even sooner.

Artificial intelligence is taking the world by storm, especially in the business world. If you want to explore this fast-moving market and try building your own apps which leverage artificial intelligence techniques and services while retaining maximum productivity then read more on what Embarcadero can do for your AI aspirations.


What's New for RAD Studio 11