← Back to the archives
AI/ML · Tech Politics

'SI' Is Not Accurate

An opinion piece arguing that renaming Artificial Intelligence (AI) to 'Super Intelligence' (SI) is a technically inaccurate rebranding

facepalm

Warning

This is an opinion post that is going to be political.

What…?

There are many pointless and nonsensical things that this current administration has done or is doing, but one of the recent ones bothers me because I feel it doesn’t make sense on a technical level.

Trump’s administration has decided to change AI, or Artificial Intelligence, to SI (“Super Intelligence”). To explain my reasoning here, let’s go deep into the world of AI.

Artificial Intelligence

Let me get this out of the way: When the media or your friends talk about AI, they only think of ChatGPT or Gemini. While that thought is technically correct, many people think AI is exactly ChatGPT or Gemini.

This is why I want to educate people on what AI really is and what something like ChatGPT would be REALLY called.

Artificial Intelligence is not new. It is very old… 1950s old. Alan Turing was the best-known figure who asked a question that was important for the creation of AI – could a machine think? I won’t go into who created AI and what the first AI was/did… for this post, I just want people to know that AI isn’t new.

So, what exactly is AI? While the definition of AI is still highly debated, one common definition is code that is able to perform a task that a human can do or code that can react to its environment.

When going by this definition, there is an example that most people would recognize… some may have interacted with it. Let me give a hint:

pong

Recognize this? Yes, this is the game Pong, the game that was released by Atari in 1972.

So, where does AI come into this? This is a game where two kids would just play against each other. How is that AI? Well, later on, a single player mode was added to Pong. And what controlled the other paddle in this mode? The computer! Or really the code. Moving the paddle to hit the ball is a task a human can do… and reacting to the ball coming its way and hitting it because of that? Sounds like reacting to its environment to me. See where I am getting at?

To a bunch of experts, the computer you play against in Pong would be AI.

Machine Learning

Okay, now you see how broad AI is. So, let’s go to the next step to get closer to ChatGPT and the others by looking at one type of AI: Machine learning, or ML.

Note

You may sometimes read or hear the term AI/ML. ML is a subset of AI, so AI/ML is shorthand that covers both.

ML refers to a machine that is able to learn by itself… with the help of humans, of course. You have likely used several tools that utilize ML such as “People You May Know” recommendations, credit card fraud detection, and email spam filters.

The main part of ML is training, which is where data is fed to the machine. The machine then goes through this data to start finding things like patterns in the data. The machine will also check its guesses against the right answers and adjust its understanding based on what it gets right vs wrong.

The result of ML is that the machine (or really the code) is able to use the patterns it learned to recognize items that follow those patterns such as spam emails and “People You May Know”.

Neural Networks

Now we are in the area where ChatGPT falls: Neural network, also called neural net or NN. Loosely inspired by a brain’s neural network, researchers believed that they could make a computer function like a human.

One example of a brain’s function that researchers believed computers could do is reasoning. Instead of reasoning the way a human brain would, a computer neural network “reasons” by learning from data and making predictions.

Going deeper, one thing that can be built on a neural network is a large language model, or LLM. This is where Gemini and ChatGPT would fall because they are chatbots that are built on LLMs.

Pong’s NPC is “Super Intelligent”?

With that information in mind, let’s go back to the renaming of AI to SI. First, “superintelligence” isn’t a new word. It already has a meaning in the field: a hypothetical AI that outperforms humans at virtually every kind of thinking. We aren’t there yet. ChatGPT and Gemini may be impressive, but they still make mistakes like hallucinations. So calling them “Super Intelligent” is a stretch before we even get to the naming problem.

Second, remember how broad AI is. Even the computer player in Pong is AI. If AI gets renamed SI, then that paddle-moving code is “Super Intelligent” too. All it knows to do is move up and down and hit the ball. I wouldn’t call that “Super Intelligent”.

I get that the administration really means things like ChatGPT and not Pong, but that doesn’t make it a technically accurate rename. AI is the umbrella term for the whole field, and SI is a specific, hypothetical level of it that hasn’t even been reached yet. Swapping one for the other is rebranding, not correcting. What do you think?