Artificial Intelligence, Machine Learning and the Future of Marketing

On this week’s episode of Ask Matt, we discuss artificial intelligence in marketing technology, the misuse of the term, and how it differs from machine learning.

Transcript below:

Matt Mroczek: Do you feel like there’s a lot of confusion between the difference between artificial intelligence and machine learning and just like the general public’s view?

Matt Lo: Yes there is a huge difference because artificial intelligence is a huge concept.

And machine learning is a little…is a substantial box but not encompassing inside artificial intelligence. And then inside that box is deep learning. So imagine you see artificial intelligence then machine learning and then deep learning.

As a marketer, what do you think of A.I.?

Matt Mroczek: As a marketer I almost, I’m curious to see if the public is going to find A.I. to be like an intrusion of their privacy. I think A.I. and I think of it like from the minority report perspective where you walk into a store and it knows your name it knows your likes or dislikes and it just kind of serves you up whatever you’re looking for. And I’m wondering if how far away that is and how people are going to feel about it

Matt Lo: When it comes to privacy, what’s your stance on that?

Matt Mroczek: I think that as much as we want and want to believe that we’re in a private world…that we have control over it, the more technology we have, it’s also on the user, the more things we introduce into our lifestyle that collects our data the more we’re admitting or giving anybody behind the scenes control over that data as well.

Why do people confuse A.I. and machine learning so much…because there’s a lack of understanding or…

Matt Lo: I think it’s due to marketers just marketing A.I. to its fullest extent.

If you have any piece in your software that is learning from experience, right? The easiest form of machine learning. They are going to amplify that all the way out to A.I.

So then you have like chatbots that say they have A.I. when really humans script it and then the A.I. part they say is like which scripted version does better based off experience.

It’s a total misconception. So then when people hear like machine learning, that’s actually more of the technical term of what’s really happening.

Matt Mroczek: So are they wrong by saying it’s A.I.?

Matt Lo: No, they’re not wrong which is why they can advertise it. They’re just amplifying the truth. It’s more like a perceived value versus intrinsic value.

Matt Mroczek: What’s like an example of that, like you would see that, just like general marketing world

Matt Lo: Well in general marking world, like saying you have a self-driving car…right? Like Tesla’s car. But if you try to say…if you want to self drive from your garage to here at mHUB. That’s not going to work. You can only be self-driving parts of it.

In the next five years, based off your judgment, based off our previous conversation…

How do you think A.I. is going to help you as a marketer?

Matt Mroczek: I think what it’s going to be able to do is…it’s going to increase the level of knowledge you have in making the right decisions.

So as you start creating a product, it’s going to help figure out what the problem and the solution is.

As you start figuring that out, it’s going to start identifying what your persona is.

As it starts identifying your persona, it’s going to start…ooo…well here’s the messages and the emotional value in it.

I think once it taps into like your emotional strings that’s where marketing and A.I. is really going to take off.

What’s the simplest form of A.I. that we interact with on a daily basis without even thinking about it?

Matt Lo: Google. Google search typeahead is all machine learning driven.

Alright, so I’m a big Fortnight player and in Fortnight they have these challenges and most of the time if you don’t play a lot…like if you don’t play like 12 hours a day you really don’t have a clue on how to do this challenge so you just Google them real quick.

A lot of them are really generic, like golden balloons…right? No context whatsoever. You just type in golden balloons, in Google typeahead based on your preferences and based on everyone else’s preferences and based on the relevancy the first typeahead result that comes up is Fortnight golden balloons.

Matt Mroczek: Does that creep you out or does it…just make life better?

Matt Lo: It makes life better…and that’s a value proposition for yeah it’s supposed to make things better.

How do you think A.I. can improve this video series?

Matt Mroczek: I think A.I. would be able to make decisions over a lot of different factors.

If we’re the right people, how our voice sounds, what the production quality looks like, what the audio looks like…is it hitting the audience. I think it’s gonna take in way more than what like a YouTube analytics is going to provide into much more of the audience and the connection in between.

What would web sites look like in five years with better data, with machine learning, with A.I….how will they change and what will the experience be for the user?

Matt Lo: I don’t think they’ll change that much in terms of visual, what will change is contextual.

So there are already services that do this right now but imagine a video. You’re watching a promotion video…and instead of just talking about the product…it’s actually personalizing that video to you.

So as an example, if you were to pretend…if you’re an agency, you help solve PPC problems. Something you do, you run digital ads.

On my video, instead of showing a generic video of showing how ChipBot works, I’m going to show ChipBot working in a landing page connected to PPC…connected to a Google, Google AdWords or Facebook ads.

And it’s going to be a quick video on how this helps you in that particular sense.

Matt Mroczek: So just more personalization?

Matt Lo: Deep personalization. And we’re talking about things that can be created on the fly.

So that video I just mentioned? That wasn’t pre-created, that was artificially created video that happened on the fly.

Why do you think people think A.I. is a killer robot or this thing that’s going to take over the world?

Matt Mroczek: Definitely just a lack of understanding.

And I think a minimum, form factor, you know how would a robot ever be constructed for any of that piece but even beyond that.

I don’t…I think that there should be like a Spider-Man quote, with great power comes great responsibility. I think as we see more companies evolving over the year, you will see a lot of legislation.

And right now legislation is always a couple of years behind technology and I think you’ll start seeing newer leaders in our democracy and across the nation taking a much more forward-thinking approach to it.

Do you…could you see…black hat techniques or like hackers using A.I. to then penetrate companies as well?

Matt Lo: I…well look right now…we have…have you ever heard of DeepFace?

Matt Mroczek: No.

Matt Lo: It’s where…machine learning can put someone else’s faces on top of the video and you can’t tell the difference. So imagine my face on your face and your face, and mine and you couldn’t tell.

Matt Mroczek: I think Jordan Peele did this with Obama. Barack Obama didn’t he?

Matt Lo: Yeah, he did. That was on CNN.

Matt Mroczek: Yeah.

Matt Lo: So if you can do that, maliciously through video, imagine what you can do at a company.

Imagine what you can do in terms of finding zero data bugs. Which are bugs that no ones reported, but they’re bugs in the system.

Matt Mroczek: Is artificial intelligence the future of marketing?

Matt Lo: No…I mean marketing, as I tell everyone, is sort of like a blend between art and science. So the science part will definitely be amplified with A.I., but we’re far from that. 

You need a standardized data input from all your sources…right..from PPC campaigns to modern social chatter…to senior conversion rates…to senior churn rates.

That all needs to go to a central source and then once you’re there you need the right models to do something with that data and then those models need to be tested.

Then you come up with these indicators or these…these predictors that you need to then find action items from them.

So we are really far away from that. I don’t know one company that at least like a startup or small business could use. That could take all that data and make meaningful Insights from them.

Matt Mroczek: Whenever that time period comes, do you think that’s going to be enterprise level only who can afford this or do you think it will be wide-stream?

Matt Lo: Personally, I think it’s going to actually happen through an open standards community. Because right now all the data is in silos.

You want your Google Analytics data? It’s in Google’s warehouse.

You want Intercom’s data from your chat history, that’s in Intercom silo.

You want your NPS scores, that’s in your feedback silo.

You want your Facebook PPC’s, that’s in Facebook’s warehouse.

You need a unified standard of talking across different channels. So then you can build software on top of that standard

Matt Mroczek: Who do you think is going to lead the charge to do that?

Matt Lo: It’s going to be…a company with a lot of money with that…that can fund that initiative.

It could be a nonprofit organization that starts out and starts fundraising.

But I think it will actually spawn out of..one of the tech companies just because they have the resources and capital.