Mind Readings: Automation is not AI

In this episode of Mind Readings, some clarification. A lot of people seem to be mixing up automation and AI. Automation is getting a machine to do an existing process automatically, hence automation. Artificial intelligence is getting machines to replicate human intelligence skills such as vision or natural language processing or uh you know specific types of mathematics like regression analysis. The two are not the same thing.

Uh even though, particularly in like mainstream news sources, the two kind of get conflated. Uh they kind of get mixed together. They're not the same thing. Automation is any type of repeatable process, right? That's the heart of automation.

You can create some amazing automation with zero artificial intelligence, right? For the last 40 or 50 years, we've had computers, sure, automation, um, doing crazy good stuff to make our lives easier, right? An Excel spreadsheet is automation, and it is incredibly good at what it does. It makes our lives easier. It makes making our sure our budgets are correct a whole lot easier.

It has improved business. A car is a form of automation, right? Um, it is the systems in place in a car can help automate the process of transportation. Um, going from walking everywhere, right? To or running or riding a horse, to a mechanical substitute that delivers a repeatable result.

That's automation. We have had automation in every industry. Go to um a car manufacturer and look at the assembly line. Even go to like I was in um gosh, where was that? Fort Wayne, Indiana, uh company called Seckler's Pickles, old, old school manufacturing company.

Um, they make pickles, a lot of different flavors of pickles, like 70 some odd flavors of pickles. And there's the assembly line, and there's there's people you know packing pickles in the jar in each jar. There's this one woman who sits at the end of the line and pushes the last pickle in the jar before the jar gets sealed and stuff like that. Uh, I don't know if that they're still there. That was a long time ago.

That might have been fully made robotic now. But that's automation, right? An assembly line, a conveyor belt. That's automation, making a task run by machine instead of by a human. AI is about giving machines in large amounts of data of some kind and saying, analyze this, categorize this, or create more of this.

What separates AI from regular automation is the learning part. That's why a lot of folks in the field use the term machine learning rather than AI, because machine learning is explicit in saying there has to be a machine that is learning on its own from the information it's given. That's what separates out your spreadsheet from a machine learning tool, right? Your spreadsheet, when you're using it, you are doing the work. You are doing the thinking.

Um, the spreadsheet's making it easier. When you're talking about machine learning, it is doing the learning, it is doing the thinking. And you may have some uh automated portions of that, but it's human portions of that, but for the most part, the machine is learning from the data and writing its own code, right? That is what separates machine learning from regular automation. The machines have to be learning from some data in order for it to be machine learning.

So why does this matter? Why do we care? Well, part of it's because we want to make sure we're using the right terminology, particularly if we're dealing with vendors uh or trying to build a strategy, some kind of business strategy or marketing strategy. We we know we need to know the difference between these two things. But the bigger picture perspective is that it helps us understand what kind of problem each area solves.

Automation solves the issue of having repeatable processes, right? If you've got a process that a human does, it's gonna have more variability in it because that's human nature, right? The last pickle you push in is not going to necessarily be exactly the same in every single uh jar. The next burger off the line when a human's cooking it may not be the same as the previous one. When you automate, you get the repeated process.

This is, you know, the the next burger should look exactly the same as the last burger, should look exactly the same as the a burger from a you know a thousand burgers ago, they should all look exactly the same in automation. And so if we have a problem that requires a repeated process with a repeated outcome, we need automation. If we have a problem where there's so much data that we can't learn from it in a in a timely enough fashion, that is an AI problem, right? So if you're dealing with something that fits in a spreadsheet, it's probably not an AI problem because let's face it, that's not a whole lot of data. And you can use regular tools, regular automation tools to probably get to the answers you're looking for.

If on the other hand you have so much data that there's no reasonable way that you could process the data and get a result, that's a candidate for machine learning when the machine can learn from it instead. Say you know, you want to summarize 70,000 articles in the next hour. Guess what? That ain't happening. It doesn't matter how fast a reader you are.

That's just not happening by you. But that is a hundred percent something that would be appropriate for machine learning, a machine learning task. So which one you choose depends on the kind of problem you have. Be very clear about what kind of problem you have. Do you have a problem that just needs a repeatable process with a predictable outcome?

That's automation. Do you have a problem that you have too much data and not enough resources or time to troll through it? That's a machine learning problem. That's the importance of this. So uh if you've got questions about automation and AI, let me know.

Love to hear from you in the comments. Otherwise, thanks for tuning in. I'll talk to you next time. Take care. If you like this video, go ahead and hit that subscribe button.