AI Terms You Should Know

Ok, AI is a thing and you want to know what people are talking about when they start chucking out the acronyms. We got you.

We'll also just say that everybody's got to start somewhere and this stuff is complicated. So don't feel like a dummy.

Even at Cleo, an AI company that’s been in the game for 6 years, there’s a few of us who need a refresher from time to time. And by a few of us, I mean the marketing team, but we move…

Here’s the AI terms you should know.

‍

AI terms you should know

(In alpahetbical order because...words 🤷)

‍

Artificial Intelligence (AI): A branch of computer science whose goal is to create machines that can mimic human intelligence. This includes stuff like understanding language, learning based on previous experiences, and solving problems. 

Part of that problem solving ability is making inferences and solving problems that don’t explicitly appear in the data they’ve been trained on. 

For example: Midjourney (the AI that generates all those fun images) can create new pictures that are works of fiction. Like a billionaire doing their tax returns. 

‍

AI assistant: A piece of software that uses AI magic or techniques (like machine learning) to help you get sh*t done. You’re usually able to communicate with them through text or voice. Alexa, Siri, and Cleo are all in this category.

If you want to know more about How Cleo Uses AI, we got you covered.

You could also watch this video.

Artificial General Intelligent (AGI): This is usually what everybody thinks of when they first hear about AI. A machine that can do everything a human does, but better. That’s a decent definition,but we’d like to note, it’s very theoretical and isn’t likely to happen any time soon. So no need to run off and join the resistance, yet.

The word general in AGI refers to the ability to be good at a range of human behaviors or skills, from reading emotions to reasoning, to being able to transmit what you’ve learned to something else. 

The opposite of general is narrow intelligent, which implies being good at fewer tasks, or excellent in a certain niche. A couple of years back, most AI used to be considered to have narrow intelligence, but the lines are getting a bit blurred now.

For example, ChatGPT can write you a poem or tell you how to fix your car. So in some ways, it’s more generalized then narrow, but – typically when you hear the phrase AGI they mean the super-AI that’s just like a human.

‍

Algorithm: The word algorithm literally means a set of instructions for a computer to follow to perform a certain task.

Let's look at social media for example.The main algos there are a set of rules that determine what the computer should show next on your feed, based on what you’ve interacted with in the past. How they're used and what they're designed to do can be controversial, but at the end of the day, algos are just intructions for a machine to follow.

‍

Alignment: Alignment is about making sure AIs remain human-friendly. In other words, as these machines become more and more intelligent and start thinking and acting independently, it makes sense that they are aligned to human values and objectives. 

Again we tend to jump to doomsday, but it’s way more practical. Imagine an AI robot that helped you in the kitchen. It’d be great if it stopped chopping carrots when a person came near the cutting board. 

If you still can’t shake the idea of killer robots and it’s creeping you out, then we’d like to paraphrase the head of machine learning at Meta, Yann LeCunn, and point out that intelligence and the desire to dominate are two completely different things.

In fact, a really smart entity would likely choose co-existence as the optimal path 💙

‍

Chatbots: You know what a chatbot is. But what you might not know, is that most chatbots can’t be considered AI. They’re usually “rule-based”, meaning they can only answer a few specific things that are pre-determined. This is your classic webpage support bot that makes you want to scream into the void. 

The next level of chatbot (machine learning based) use AI techniques to better understand what you’re saying and get you the correct info faster. But they’re usually not generating new content on the fly.

Some chatbots though, like ya girl Cleo, can be considered AI. They’re more sophisticated and can offer conversational capabilities that can sometimes rival human performance in specific domains. 

‍

Co-pilot: A catchy name for an AI assistant. 

‍

Deep Learning: This is a subset of machine learning but focuses on using deep neural networks, which actually get their name from what they're based on: neurons in the brain. The term "deep" refers to the number of layers in the neural network. More layers allow the network to learn more complex things.

Which leads us to the next definition…

‍

Deep Neural Network (DNN):  The easiest way to understand a neural network is to imagine data from various inputs (neurons) coming into centralized hubs (nodes) and being processed according to a set of rules, which leads to a different output.

They can be layered on each other as well. So the output of one neural network becomes the input for the next layer.

It’s like when your eyes send the information about what they’re seeing to your brain. The brain then puts the pieces together to make the image. A different part of the brain takes that output (the image) and tell you what you’re seeing and how you should react. 

A few neural networks are not “AI’ themselves, but a big building block of it. It’s when you start to stack them together that you get a Deep Neural Network which can do the whole deep learning / AI thing.

‍

Large Language Model (LLM): A type of artificial intelligence model that’s good at understanding and generating human-like text. This is your ChatGPTs and Claude’s of the world. 

They’re good at things like translation, content generation, question answering, and summarizing long pieces of text into shorter version. And of course, they make for great chatbots. 

They’re called large because they contain billions of variables they used to make predictions of decisions. Yeah, we said billions. 

‍

Machine Learning (ML): Machine Learning is a subset of AI. It’s about getting computers to perform a task without needing explicit instructions. In other words, it’s like the machine “learns” from the data. 

‍

Natural Language Processing (NLP):  A field within machine learning focused on getting machines to understand language.

So for example, if you tell Cleo you’re stressed about money, she’ll sense that you’re probably not feeling great and offer some helpful suggestions. 

‍

Singularity: As it relates to AI and technology, the singularity is the theoretical moment when we have AGI (the AI that can do any human thing like a genius) and the result is an exponential explosion of technological advancements that fundamentally affects human civilization.

Again this is largely theoretical. Some scientists think it’s not many years away. Some think it will never happen at all. 

‍‍

If you want to start using AI for free...

We’d obviously suggest giving Cleo a try. She’s the first AI assistant that’s dedicated to helping you do the money thing better. 

It’s also a good time, chatting to her is an experience you won’t find anywhere else. 

Until next time. Big love.

‍

‍

‍

‍