So... What Exactly is AI?

By James

It feels like a lifetime ago when AI simply didn't exist in our lives, when we relied solely on search engines to find our answers or even genuinely coded an entire file ourselves. However, only two years ago, something changed all of that. A few models relating to language, known as Large Language Models (LLMs), were formerly irrelevant for the most part, only used within academia and the often unnoticed internal components of software. Yet ChatGPT, a product of those models suddenly became relevant, gaining tens of millions of users within the weeks that it launched, placing the world's attention on what had formerly been an obscurity, ushering in a new period of technological development within AI.

An Anecdote

I must admit, as well, that if it weren't for the rise of ChatGPT and similar products, I would not be writing this article right now, I would not have even gained interest in the field of AI and instead, just carried on spending all my free time playing random video games. As a side note, however, I would have quite enjoyed Tears of The Kingdom, but long story short, I got into AI and am now devoting my free time to related interests.

What AI Is

Amidst the rise and popularity of AI, though, one question remains quite apparent, but no one ever tends to ask it: What exactly is AI?

To answer this, we must look at the causes behind ChatGPT's popularity and notoriety. Search engines used to be the primary method of getting answers to all sorts of questions. You had to look up what you were wondering and browse through a few web pages to see if one of them had relevant information. However, with AI, an answer can be found without needing to browse through sometimes relevant, sometimes irrelevant websites, presenting information in an understandable, accurate--and mostly concise format. The human-like, conversational aspect of ChatGPT also made it all the more interesting to many, posing itself as essentially a person who can answer any question you have, listen to you vent out stress, or even provide advice on life's various issues.

We can think of ChatGPT, primarily the model behind it, as essentially a human mind or human intelligence synthesized for a particular usage, hence the term Artificial Intelligence. In this case, as a conversational and accurate chatbot. AI, as a whole, can also fall within a tent of synthesized human intelligence tailored for a specific usage from the Large Language Models behind ChatGPT to car autopilots.

Each model learns the skills needed to perform well in their intended usage and outputs responses based on data, otherwise a prompt or question. Some are also given various "senses," in a way, possessing the ability to see, hear, and even taste, touch, or smell. Some models, such as LLMs for the most part, can only "hear," essentially processing only linguistic or text inputs. Others can only "see," taking either images or videos as inputs. However, there are a few hybrid or multimodal models capable of processing inputs with more than one "sense." For example, some LLMs also utilize images combined with text as inputs to improve the accuracy and quality of their responses. There is a lot more to AI than its foundations in human intelligence, but to put it simply, AI itself is a synthesized form of that intelligence, tailored to specific usages.