Artificial intelligence (AI) has been powering the tools we use in our everyday lives for decades now. And every time powerful advancements in AI are released, conversations around the good, the bad, and the ugly of AI inevitably dominate the headlines.
Amid all the noise, I've noticed people throwing around technical buzzwords like "AI" and "automation" and using them interchangeably. Only, they're not the same thing.
Here, I'll break down the key points you need to know about the differences between AI and automation. And, more importantly, how to get the most out of both (hint: it involves human brain power).
Table of contents:
What is automation?
Automation is simply setting something up to run automatically. The heart of any workflow automation boils down to a simple command: "When this happens, do that." For example, when someone fills out a form on your website, then automatically add that contact to your email list.
What is automation used for?
Automation is great for replacing repetitive or mundane tasks, which is why the ability to automate workflows is baked into a lot of the apps you already use. Take a scheduling app like Calendly, for example. Instead of manually sending meeting reminders to attendees prior to every meeting, Calendly does it automatically. You're not limited to only automating workflows within a given app, either. Some apps have native integrations that let you automate across apps. Or there are automation tools, like Zapier, that let you do the same across thousands of apps.
When you automate these types of tasks, it ensures consistency, reduces the risk of error, and frees you up for more high-value tasks.
While automation is really good at following a predetermined path (or set of rules), it falls short when an action along the path requires interpreting data and making a decision before it can proceed. That's where artificial intelligence comes in.
What is artificial intelligence?
There are multiple definitions of artificial intelligence, ranging from "a poor choice of words in 1954" to "machines that can learn, reason, and act for themselves." For the purposes of contrasting AI against automation (and, later on, understanding how they work together), I'm using the more nebulous definition of machines that can, to some degree or another, "think."
AI's ability to "think" comes from machine learning—a subfield of artificial intelligence that enables a system to analyze massive datasets, learn from that data, and then make decisions based on it. (This is a gross oversimplification, but you get the idea. For more details, here's a basic guide to AI.)
What is AI used for?
AI is already baked into a lot of services you probably use every day, including:
Recommendation algorithms on Amazon, Netflix, and other websites
Spam filters in Gmail and other email apps
Fraud detection for your credit card, bank, and other financial services
But since ChatGPT came on the scene and spawned thousands of new AI-powered tools, AI is also changing the way we work. For example:
AI chatbots like ChatGPT and Gemini help you summarize lengthy articles, analyze data, and think outside the box.
AI text generators like Jasper and Notion AI can act as your writing assistant, generating entire article outlines and short-form copy in a matter of seconds.
AI image generators like DALL·E 3 and Midjourney can help your business with things like creating images for social media posts, building instant presentations, and sketching storyboards.
And that's just generative AI—there are also tools like predictive analytics software that can analyze massive amounts of data to help you make decisions. The caveat with AI is that it's highly dependent on human prompting and accurate data—the output will only be as good as the input you feed it with.
How AI and automation work together
Some of your daily workflows are probably straightforward—for example, when you react to a Slack message with a specific emoji, then it automatically gets added to your to-do list app. But what happens if you want to add a more complex step to that workflow, like labeling the task as high or low priority depending on the context? Automation on its own can't handle that step.
To build a truly powerful automated workflow—one that addresses these kinds of gaps—you need to add AI to the mix.
Here's an example of AI and automation working together in one of my own Zaps (what Zapier calls their automated workflows).
In essence, this is what's happening: whenever an article I've published appears in my RSS feed, Zapier adds it to Airtable as a new record (this is the automation part). But before that record is created, ChatGPT scans the article and decides what category it belongs to based on a detailed prompt I previously fed the chatbot (this is the AI part).
Now, when I look at my list of published works in Airtable, I can see details like the name of the article, the site it was published on, the URL, and the category it falls under—all without having to lift a finger.
In this example, it's the blend of AI and Zapier's deterministic workflow engine (a system that executes tasks in a predictable manner) that reliably processes the workflow the same way I would if I had to do it myself.
Where do humans fit into all of this?
Whenever someone cries "the robots are coming for our jobs," I think about ATMs (my background is in banking, so this isn't completely out of left field). In the 1970s, the increase in ATM use sparked concerns that this technology would eventually replace human bank tellers. But that's not what happened. Instead, ATMs freed up tellers to do higher value work, like answering more complex customer questions.
Because while AI and automation have the power to take on a lot of tasks—especially the monotonous ones—humans are still needed for work that:
Is unique
Requires a point of view
Requires critical thinking or reasoning
Builds on relationships
Even so, this doesn't mean you can't leverage AI and automation for these types of tasks. Based on my experience, there are plenty of workflows that would benefit from AI, automation, and good old-fashioned human brain power.
Take an automated approval workflow, for example. With Zapier, you can use AI to automatically mark a request as approved (or rejected) based on prompts you've created, but then give you (the human) final say over every request. Or you can use the AI to approve or reject straightforward requests and flag outliers for human review.
AI vs. automation vs. human power: they're not mutually exclusive
We've barely scratched the surface of what's possible when you apply AI, automation, and human power to your workflows. But what I hope my examples emphasize is this: these exciting technological advancements don't pose an either/or dilemma—it's not AI or automation or human power. You can use them together, and, in fact, it's better if you do.
Related reading:
