WILL AI REPLACE MARKETERS?

Summary:

With rapid advances in AI and machine learning technology, many people are concerned about what this technology will mean for future employment. Will machines replace humans in the workplace? Will AI replace the need for human intellect and decision making? Stephen Yu, Chief Product Officer at BuyerGenomics, discusses the difference between AI and machine learning, and what these technologies will really look like in the workplace.

Below is a lightly edited transcript of Episode 28 of the Inevitable Success Podcast with Damian and special guest Stephen Yu. (Listen Here)

Transcript:

Damian: You recently wrote an article that was published on Target Marketing Magazine entitled, Replacing Unskilled Data Marketers With AI, and while I think you’re being a little provocative with that, I also think there could be some truth there. Tell us about that.

Stephen: Actually, the essence of the article is that a lot of marketers embrace words like AI and machine learning — by the way, AI and machine learning are two different things, but people use those words interchangeably. For users, I think it’s OK, since one is an extension of the other, really.

Now, what’s interesting is those people who embrace it really freely, they think the machine will just show up, do all the work, tell you what to do, and just shut up. There are no geeky questions or annoying follow-ups — but this is not the reality. You have to be either a person who gives purpose to those machines or a person who knows how to wield those machines. You can’t just turn the machine on and call it a day. It is really hard for a machine to mimic human decision or human behavior.

So, Rule #1: Machine learning is not doing something different, it’s just doing the things that humans know how to do already, but more quickly, automatically, and on a mass scale.

These machines are like a child, or even a small dog. That’s how they learn. If you want the machine to start learning, to get better and better and better, you have to set it up that way. You have to give it a purpose. You have to train these machines, so people who know how to do these things will always have a job. But in the case of people who aren’t executives with control over these things, it’s going to replace those people’s jobs.

Damian: It’s funny, I just listened to a TED Talk about this topic that was really interesting. They said that historically, whenever new technologies come out, there’s always been an increase in jobs. They said that’s actually been true for every worker in America except one: the horse. There were millions of full-time, working horses in the United States alone that went to work every day, and when the automobile came out, they suddenly became irrelevant. I thought it was a great metaphor for this.

Stephen: That is so true. And we further humiliate those horses by measuring the engine power in horsepower. We say, “This engine will replace 350 horses,” and I think, really? Do we really have to count horses here? It’s like saying a machine represents the collective power of 10,000 brians.

Damian: Wow, maybe you’re on to something. We can start to quote features in “marketerspower.”

Stephen: There you go.

Damian: We digress.

Stephen: You know it’s all related, though. Think of it in the context of revolutionary change. What does this all mean? Even the agricultural revolution was good and bad. I was reading a really interesting book called Homo Sapiens that talks about how we came to be the humans we are today. It discusses how the success rate of a species depends on how you measure that success. For example, are we a really successful species just because there are a lot of us on Earth? Well, if that’s the criteria, chickens are more successful than humans. But do we call them “successful?” Not really, right, because we raise chickens to eat. Not a good thing for the chicken.

Now, the agricultural revolution that I’m talking about, this bound humans to the ground. Our bodies are not even designed for it, but we do this back-bending work. So is that good or bad? Well, it depends on who you are. If you own a lot of land that you rent out and collect tax on, then it’s a very good thing. It’s great if you’re running a country. If you’re a farmer, you were happier when you were a hunter-gatherer, when you were only responsible for obtaining what you could eat, and you were probably healthier, too. And you look at these marathon runners, and you think, this is what our ancestors looked like. And it’s what someone in prime condition looks like, as compared to someone sitting in front of the TV set all weekend. So, I digress, really, but the point is, there are good and bad aspects to any change that is revolutionary.

Damian: This is almost a philosophical topic. It’s easy to go there.

Stephen: It is. So let’s talk about how we survive as marketers or any decision makers in the age of machines and the age of abandoned and ubiquitous technologies. Let’s make another analogy: I read an article about a program that writes music all by itself. Now, if the goal is to just create a lot of fillers and play them in an elevator, then it could work, and the machine would do a fine job, because you don’t really pay attention to it. But would you write music intended to move hundreds of people, in the theatre, live, that way? No.

What I’m trying to say is that it doesn’t replace everything. There are some human elements that have to be kept, and that applies to the world of marketing. We talked about personalization a couple episodes ago. Personalization is mimicking personal touch, and that’s what machines do. They get better and better, and it gets harder and harder to tell the difference between a machine and a human. But you know what, sooner or later, we know it’s a machine, because it’s too formulaic and too predictable to be a real human being.

Down the line, machines may be able to replace all of us, but the point is, should they? Even if you have machines like that, who gives them the purpose? We have to be the decision makers. We have to be the people who wield the machines for our benefit, and to do that, we have to speak the language of logic. And you can’t just sign off old-fashioned modeling, either. The neural network was created years ago, for example, but only now do we have the computing power to harness the power of it. So how do you embrace that and not embrace the old-fashioned modeling, because they’re really the same thing. That kind of thinking is how you become irrelevant. You have to break down the whole thing in a logical fashion to see if the machine or human-made models, or whatever, is beneficial to your next move in marketing. Only then do you do it.

So on the one hand, people are overly concerned and scared that they’re going to become irrelevant and lose their jobs, and that machines will revolt and kill everybody like the Terminator. Or, on the other hand, people think they can just turn on the machine and go home while it does everything for them. I don’t think so, it doesn’t work that way.

Damian: I think, no matter what business or brand you’re talking about, they’ve never done everything they physically can to maximize their marketing. If you go by that logic alone, there’s always something more to be done to improve performance. But you don’t do it all, because of diminishing returns, or you don’t know exactly what to do. As technology advances to a certain level, it allows you to get closer and closer to expanding the things that you’re doing in your marketing. If you take a lesson from the horse, the horse went away because it couldn’t adapt to learn a new trick. With humans, that’s one of our key skills. So if we embrace the fact that there’s always something more that can be done, and that this just lowers the cost or increases your visibility to know what to do next, then it can be very powerful for the people that take that point of view.

Stephen: That’s exactly the point. People who take that point of view, not people who worry they will be replaced. What is a machine anyway? Forget about the mission of learning for a second, and let’s go back to the machine itself. Let’s talk about the industrial revolution. You take 20 employees hired by a small manufacturer and you replace them with a single machine capable of doing the work of a hundred people. That was scary. Did people starve? Some people did, people who did not learn new skills to wield the machine. Or maybe they went back to the farm. We all have to eat, right?

They say that the service industry, the human contact oriented industry, will survive all this, and that’s fine. But again, let’s go back to what to do first. You pointed out an interesting fact, which is that you have to decide what to do with these things. You cannot just say the machine will figure out what to do on its own. We have to prioritize what is more important. That prioritization is all about analytics — numbers and data. You cannot just rely on gut feelings anymore. You have to have some empirical evidence that if you do this, it will make you more money. Again, it’s about doing what you know how to do, but doing it faster, better and more efficiently.

Let’s say it’s a known operational problem that you have too many customer service calls. Wouldn’t it be nice if some machine could categorize what the problem is and prepare a script for the human being who answers the call? This would be much more efficient.

But someone has to analyze each step in the process to determine which parts will be automated. All of these steps are human work. To break it down, you need a lot of knowledge in machine learning, and computing and data, not to mention being familiar with the whole entire process. You don’t want to just buy some technology and call it a day. People will not just line up at your door because you bought a new technology. The improvement is incremental, and machine learning is the same way. It can be used for revenue increase, but mostly for cost reduction by doing things faster. Then you have to decide what should be automated, how it should be automated, what human interaction there should be, and how do we design these things. That’s the human work. If you know how to do that, you’re fine.

Damian: One metric of intelligence is that you are able to solve a problem you’ve never seen before. As AI gets more sophisticated, and machines get better at solving problems they’ve never seen before, I think that what you could work with could change. But who knows how long this is going to take and, more importantly, when it will become affordable.  

Stephen: I think the affordable part is there, although it’s all cloud-based, machine learning algorithms that are time-tested, which means they’re clunky and have been out there for a long time. But they are available, and you can wield them if you know exactly what to accept as what I call “cost of being wrong.” But that’s okay, you can do that. You don’t have to jump into some million dollar solution right now. If you don’t have a clear goal, don’t spend a single dime, because it’s not some magic wand that will free you from having to think about anything.

Damian: Certainly not. I think we’ll continue this conversation on in later episodes where we can get more into the details of what AI is good at. That was great, and I encourage everybody to read the article, Replacing Unskilled Data Marketers With AI, by Stephen Yu. Thanks again.

Damian: Thanks so much. If you enjoyed today’s episode, we ask you to please leave a rating and write a review. Or, better yet, share with another marketer. If you’d like to speak to someone about any topics covered in today’s episode, please visit buyergenomics.com and start a chat with the team today.

2018-11-14T21:44:28+00:00