What Does it Mean to "Work with AI"?

Everyone keeps telling you that you need to learn AI if you want to future-proof your career. And sure, that sounds like good advice but what does “working with AI” actually mean?

• Are we talking about using ChatGPT to help you write emails, analyze information, or get through your current workload faster?

• Or does it mean learning how to connect AI to other tools and automate entire parts of a business?

• Maybe it means writing code and adding AI features to apps

• Or perhaps you want to go much deeper and actually build, train, and improve the AI models themselves

• You could even end up working on specialized AI systems used in things like healthcare, fraud detection, robotics, or self-driving cars

The slightly confusing answer is that all of these count as working with AI. 

But the good news is that you don't need to learn all of them.

There are different levels of working with AI, requiring very different skills. Some don't require you to change careers or even know how to code. Others can lead you into entirely new careers as an AI Developer, AI Engineer, Machine Learning Engineer, or Data Scientist.

So rather than telling you to “learn AI,” let's make this much more practical.

In this article I'll show you the main ways people actually work with AI, what that work looks like in the real world, and the skills you would need to get started, as well as resources so you can start learning ASAP. 

That way, you can figure out which path actually makes sense for you.

So let's get into it…

Option #1. Working with AI in your day to day work

If you've ever asked ChatGPT or Claude to help you solve something at work, then technically you've already worked with AI.

However, there's a BIG difference between occasionally using AI and actually being good at working with it.

For example

99% of people will ask ChatGPT a question, get an answer, maybe copy and paste it somewhere, and move on. And sure, that's useful, but you're still doing essentially the same job in the same way. You've just made one small part of it a little faster.

The bigger opportunity comes when you start looking at your work and asking “Why am I still doing this manually?”

For example

Imagine you work in operations for a company, and every Monday you have to create the same weekly report:

• You pull data from five different sources

• Add everything to a spreadsheet

• Clean it up

• Look for unusual numbers

• Compare it with previous weeks

• And then write a summary for your manager

That’s a lot of stuff, so in the past it probably took you around 5 hours. However, because you’re awesome and you've done this same report so many, you’re now far more efficient and get it done in 3 hours instead.

That's great and you've saved yourself two hours a week. The problem though is that you're still spending 3 hours every Monday doing essentially the same thing.

That's 156 hours a year! Doesn’t seem like much but you’re basically spending 1 month out of 12 just filling out a report…

Now imagine you know how to work with AI properly

You build the workflow once and then hit go each Monday.

Your AI system then:

• Pulls the data from those five sources

• Puts it into the correct format

• Checks for unusual changes

• Compares those numbers with previous weeks

• And prepares a first draft of the report

You might even have an AI agent handle several of those steps automatically every Monday without you even having to hit go. More importantly though, now your job isn't to spend three hours assembling the report. 

It's to spend 15 minutes checking the data, reviewing what the AI found, correcting anything it got wrong, and making the final judgment calls that actually require your expertise.

You've gone from doing the repetitive work to directing and reviewing it, and that's the important difference.

Working effectively with AI isn't just about getting better at prompting. It's about recognizing which parts of your work AI can help you rethink entirely.

Once you start looking at your job this way, you'll probably notice opportunities everywhere.

Reports, research, meeting notes, data analysis, customer support, documentation, emails, presentations, scheduling, competitor research, administrative work, and dozens of other repetitive tasks can potentially be sped up, partially automated, or handled differently with AI.

• You still need to understand the work

• You still need to check the output

• And you certainly shouldn't blindly trust an AI because it confidently gave you an answer.

But instead of spending most of your time doing the repetitive parts, you can spend more of it on the parts that require your experience, judgment, creativity, and decision-making. That's a valuable skill whether you're an accountant, marketer, developer, project manager, designer, salesperson, analyst, or almost anything else.

You can even stay in your existing career and simply become the person who knows how to use AI effectively within it, or you can take it further and do this as a full-time, highly paid role.

Either way, I highly recommend this as you’re jumping in point, so go head and check out our Prompt Engineering course:

Beginner

Prompt Engineering Bootcamp (Working With AI & LLMs): Zero to Mastery

32 Hours •290 Lessons

Stop memorizing random prompts. Instead, learn how Large Language Models (LLMs) actually work and how to use them effectively. This course will take you from being a complete beginner to the forefront of the AI world.

Scott Kerr

You'll learn how LLMs such as ChatGPT actually work, how to communicate with them effectively, and how to build more useful AI workflows instead of simply collecting a bunch of prompts from the internet.

Option #2. Integrating AI into systems and processes

In the last example, you used AI to improve your own workflow, but what if you wanted to do that for an entire company?

Well good news, because companies are looking for people who can examine an existing business process, figure out where AI could help, and then actually build the systems that make it happen.

For example

Imagine a large retail company with hundreds of stores. And in each of those stores, someone needs to keep an eye on stock levels across thousands of products.

It’s not their whole job, but it's definitely a time sink, because they need to know what's selling faster than expected, what might run out soon, what isn't selling at all, and when new stock needs to be ordered.

Order too late and products disappear from the shelves, but order too early or order too much, and you've got money tied up in stock sitting in a warehouse.

Traditionally, a lot of this work involves people checking reports, comparing spreadsheets, looking at previous sales, and making judgment calls several times a week.

As you can imagine, this takes a lot of time, but remember, this probably isn't their entire job. It's just one recurring process taking time away from everything else they need to do. More importantly, there’s someone doing this at every store in thousands of locations, every single week.

So what's the solution?

Well the company could bring in someone with AI engineering and automation skills to build a better system:

• Instead of employees constantly checking everything themselves, the system could monitor sales and inventory data automatically

• It could flag products that are selling unusually quickly, predict when stock is likely to run out, and recommend how much should be reordered

• The system might take into account previous sales, seasonal trends, promotions, local events, weather, or other information that could affect demand

• It could then prepare recommended orders automatically, while keeping a human involved to review and approve important decisions

This way the person who was getting stuck doing this for 8 hours a week, now they only need to spend an hour and is free to do the rest of their job.

That’s 7 hours saved every week x 52 weeks a year for 364 hours a year!

That’s already impressive, but now multiply those savings across hundreds or even thousands of locations. If a system saved just eight hours of manual work each week across 1,000 stores, that's:

7 hours × 52 weeks × 1,000 stores = 364,000 hours saved every year.

And that's why companies are willing to hire specialists who can build these systems.

The important thing to understand though with this role (and the last role) is that you're not necessarily building the underlying AI model yourself. Instead, you're taking existing models, APIs, databases, software, automation tools, and company data and connecting them together to solve a real business problem.

That's a big part of what modern AI engineering looks like.

You need more technical knowledge than someone simply using ChatGPT in their day-to-day job, because now you need to understand things like programming, APIs, databases, AI models, automation, and how different systems communicate with each other.

But you're still starting with the same basic question: “What are people doing manually that we could make faster or better with AI?”

If solving those kinds of problems sounds interesting, check out our AI Engineer roadmap:

It'll walk you through the skills you need to learn, the projects you can build to prove those skills, and how to work towards getting hired.

Option #3. Creating AI-powered products 

Instead of using AI behind the scenes to improve an internal business process, you might use it to create a feature that becomes part of the actual product.

For example

Imagine you work for a note-taking app and lots of your customers use it during meetings.

But the problem is that their notes are a mess:

• They've got half-written sentences

• Random ideas

• Voice recordings

• Decisions they need to remember

• And five different things they're supposed to do afterwards

You want your app to sort all of that out for them, but you don't need to spend years building your own AI model from scratch. So instead, you could connect your app to an existing AI model through an API.

Your customer records their meeting as normal, but now your app transcribes the recording, sends the transcript to the AI model, and gets it to identify the important information.

A few moments later, the customer might get:

• A clean meeting summary

• The important decisions that were made

• A list of action items

• Who is responsible for each task

• Any important dates or deadlines

Handy right?

You've taken an existing AI model and built something genuinely useful around it.

And that's an important thing to understand if you're considering an AI career in that you don't have to know how to build ChatGPT to build something with ChatGPT.

Think about how many apps use maps without their developers building a global mapping system from scratch, or accept payments without creating their own banking network.

AI can work in a similar way.

Companies such as OpenAI, Anthropic, Google and others build powerful models, then they allow developers access to them so they can build their own software and experiences around them.

That's opened up a huge area of development.

You could build AI-powered search, customer support tools, coding assistants, recommendation systems, document analysis tools, educational apps, research assistants, content tools, or features we haven't thought of yet.

This path is particularly interesting if you like the idea of writing code and building things people actually use.

You'll need programming and software development skills, but you don't necessarily need the advanced mathematics and machine learning knowledge required to develop the underlying models yourself.

If that sounds more like you, check out our complete AI Developer roadmap:

This will take you through the skills you need to start building AI-powered applications, create projects for your portfolio, and work towards getting hired as an AI Developer.

Option #4. Building and training the AI itself

Everything we've looked at so far involves taking AI that already exists and using it to solve a problem, but what if you want to go deeper and work on the AI itself?

Well, this is where we move into areas like machine learning, data science, and AI research, where you’ll build and train specialized models designed to solve particular problems.

For example

Think about a self-driving car.

It needs to constantly understand what's happening around it:

• Is that object ahead another car?

• Where does the road end?

• Is the traffic light red or green?

• Is that dark patch on the road just a shadow, or is there a pedestrian stepping into the street?

To get these types of results, engineers train models using huge amounts of relevant data so that they gradually become better at recognizing the patterns that matter, and the same basic idea appears everywhere.

• A bank might use machine learning to identify unusual transactions that could indicate fraud

• A healthcare company might develop a system that helps identify patterns in medical scans

• An email provider might use it to determine whether a message is probably spam

• A photo app might recognize the objects, people, or animals appearing in your photos

• A retailer might use machine learning to predict future demand

The problems are completely different, but there's a common idea behind them in that instead of explicitly programming every possible rule, you're using data to help a machine learning model learn patterns and make useful predictions.

This is a much more technical path than using ChatGPT at work or building an app around an existing AI API. You'll usually need a stronger foundation in programming, mathematics, statistics, data, and machine learning.

That means it's also likely to take longer to learn but don't let that put you off if this is the part that interests you, though.

You don't need to know any of this before you start learning. It simply means there's more groundwork to cover before you're ready to work professionally.

All our courses cover all this also. We assume you’re coming in with absolutely zero knowledge, and teach you exactly what you need to know from the ground up.

So if the idea of understanding how AI actually learns, training models with data, and building machine learning systems sounds exciting to you, check out our Machine Learning and Data Science course:

Intermediate

Complete A.I. Machine Learning and Data Science: Zero to Mastery

45 Hours •390 Lessons

Learn Machine Learning (Artificial Intelligence), Python, Data Science, Data Analysis, Tensorflow, Pandas & more. All using the latest in AI!

Andrei Neagoie&Daniel Bourke

It'll take you through the core programming, data science, and machine learning skills you need to start working towards these kinds of roles.

So how will you work with AI?

Hopefully, you now have a much clearer idea of what people actually mean when they say you should “learn AI.”

More importantly, you should have an idea of where you might want to start:

• Want to use AI to get better at your current job? Start with Prompt Engineering and learn how to use AI effectively in your existing workflows

• Want to build AI systems and automate processes for companies? Look into becoming an AI Engineer

• Want to build apps and customer-facing features using AI? AI Development is probably the path for you

• Want to work with data and train AI models to solve specific problems? Explore Machine Learning and Data Science

And if you're still not sure, that's completely fine.

Just start with Prompt Engineering and learn how to use AI properly in the work you already do, see which parts you enjoy, and go from there. At the very least, you’ll have one of the most important skills of this year under your belt, and you’ll be able to start to automate a lot of your day to day tasks.

From there you can get promoted internally, move into full-time prompt engineering, or any of these other roles.

Because fun fact? Prompting is one of the key skills that almost all employers are looking for - even if their role isn't anything to do with AI.

That’s how powerful this skill is!