Simple LLM Use Cases

AI this, AI that, it seems you can't do anything nowadays without seeing something either about AI or generated by AI.

So, what's all the fuss about? Should we be welcoming our AI overlords or organizing a resistance? Are these great tools we can harness and use? Or is this the beginning of the end?

As the last movies in Marvel's Infinity Saga were coming out, I was rewatching many of the previous movies. When I rewatched Avengers: Age of Ultron, I had a good laugh at a moment that wasn't supposed to be funny. Mid-way through the movie, the Avengers are down. Things haven't been going well for them. They find a safe place to regroup and Tony Stark (AKA Iron Man) is asked to take a look at an old tractor and see if he can get it working.

Nick Fury, the mastermind behind the formation of the Avengers, shows up and very dramatically and accusatorially says to Tony, "Artificial Intelligence." I burst out laughing at that. He was so dramatic and made it sound like such a negative thing I found it completely out of place. I laughed at that because I had been taking classes about Artificial Intelligence as part of my Master's program at the time and found the drama so uncalled for it was funny.

While there are many ethical considerations and new situations that have come up because of artificial intelligence, I do not think it is a bad thing. Like any new development or resource we have, we need to learn to manage it and use it within resonable bounds. After all, it's a tool that we are making, therefore, we are in charge of setting it's bounds and making sure it stays in its guardrails.

In this article, I hope to demistify the aspects of AI the average person needs to know about and show you how you can harness the power of AI in your daily life. While you may now feel a bit like Nick Fury and a few other Avengers and be very cautious about AI, I hope by the end of this article, you're at least willing to give some of the tools a try to see if they'd be helpful to you.

So, first of all, what is AI? Simply put, Artificial Intelligence is the ability for a computer or machine to accomplish tasks that usually require human intelligence. This can be as simple as deciding what to do based off of certain inputs, completing an IQ test, identifying objects, or reasoning through complex workflows to accomplish a task.

Okay, that's still a somewhat scientific answer. What is AI really? From my perspecive, AI is a type of tool we have at our disposal to help us more efficiently accomplish things we want or need to do. It's a great pitch-hitter when adjusting recipes, it's a competent sounding board for brainstorming, and it's especially good at doing the complex but repetitive tasks you need to do every day. It's decent at editing images and videos and can help you get any project off the ground.

Let's take a moment and talk about the tool that's been making waves since the first one came out in 2022.

Currently, the type of AI tool that's causing the most disruption is know as LLM's or Large Language Models. What are LLM's? They're programs that are meant to mimic human speech and resoning. They're trained on vast datasets and you can talk to them like you would another human.

To get a little more technical, LLM's are massive deep learning neural networks that have billions (yes, billions with a "b") of parameters. The latest models have over one trillion with some potentially exceeding ten trillion. What's that mean? That means these models can consider a large amount of data as they process your requests to them. The more parameters they have, the better their answers will be, generally speaking.

So, how does that affect you? What on earth could an LLM do for you? Well, do you need some help adjusting dinner because you're missing some ingredients? Ask an LLM. Chances are that LLM has trained on many cookbooks and blog articles about cooking. It will likely be able to inform you of valid subsitutes or adjustments you can make based on what you're trying to make. I've even asked it for entire recipes and it's done a pretty good job at that.

What else can you use an LLM for? Do you need help with an exercise plan or a meal plan? What about ideating over that novel you've been meaning to write? Do you want a sounding board for a party you're planning? Or are you curious the various costs involved with a ceratin purchase and which option might be best? Are you a new home owner and not sure if that creak in the floor is a problem? Do you need to prepare a training or a talk?

These are all simple examples of things you can chat with an LLM about. But their usefulness goes far beyond being a great stand-in for various topics you may not know a lot about. They're also good at writing code, summarizing text, building rough drafts for projects, or even providing feedback on your work.

The more advanced models, with the help of other tools you can connect to them, can even be a personal assistant to you. You can describe a problem you have (eg. I have a large catering order on Friday), any actions you need to take (eg. I need to make sure I have enough ingredients on-hand to fulfill that order in addition to my regular business), and ask the LLM to do something for you (eg. Please review our inventory. If we have less than X of ingredient Y, please order more so we have at least Z). The LLM will then use the models and tools at its disposal (staying within any guardrails it has) to reason about and accomplish the task.

That example is at the forefront of what LLM's are capable of today. When acting like this, we commonly refer to them as agents because they have a ceratin measure of agency and autonomy to reason about and take action in our behalf. (On the technical side, it's important to configure these agents with guardrails so they stay within any ethical or business constraints you have.)

LLM's are very powerful tools and we're only getting started there. The recent development of the Model Context Protocol is an attempt to standardize how your LLM uses and talks with other tools such as websites, other accounts, or really anything you want your model to interact with. As we add more and more tools to LLM's, your models with be able to accomplish more and more tasks on you behalf. Imagine one day being able to have your AI assistant take care of your bills or manage your subscriptions, or even take care of the one-off financial things that come up throughout the year. Or even having an LLM that can start cooking dinner for you while you're on your way home from work. That may sound futuristic, but it's entirely within the realm of possibilities given the current abilities of an LLM. The main thing lacking here is MCP enabled kitchen appliances and, well, a way to get the ingredients to said appliances.

I hope you've learned something about LLM's today. If you have, please share it in a comment! Do you currently use LLM's? If so, how often do you use them? What do you use them for? What have you found makes them more useful for you? If you don't use LLM's, what's holding you back? Are there things you want to learn more about to help you use them? Please leave your thoughts below!