Artificial Intelligence is the most groundbreaking field of science today. Lately, news about new applications for this technology, as well as AI startups getting massive funding, seems to be ubiquitous.
But the news usually comes from industries like Healthcare, Automotive, or Financial Services. We know that self-driving cars are coming, but what about robot marketers?
In this article, we’ll see what AI is, and how it fits into various things that digital marketers do on a daily basis – copywriting, campaign planning, and content creation.
Flash course on AI
The most basic definition of Artificial Intelligence comes from Minsky and McCarthy, pioneers in this field. In the 1950s, they described AI as programs performing tasks which, if a human carried them out, we would say they had to apply intelligence to perform them.
It’s a broad definition, but it means that AI is a technology that has human traits:
* Ability to manipulate and move objects
Overall, AI is a branch of computer science that deals with creating computer systems that can accomplish human tasks with human-like (or better) efficiency.
In the past 10 years, investment in AI startups has been growing steadily. Let’s take a look at a few notable examples just from this year:
* $620 million raised by SenseTime Group, Chinese facial recognition company, making it the world’s most valuable IA unicorn at a $4.5 billion valuation
* $153 million raised by UiPath, enterprise robotic process automation software company, which went from less than 100 to over 700 customers in 2017 alone
* $102 million raised by Pony.ai, an autonomous driving platform developer in China
* $10 million raised by Drishti, company specialized in using AI to digitize human actions on assembly lines
* $12 million raised by Alloy.ai, a company developing a digital supply chain platform collecting scrambled data to uncover additional revenue opportunities
* $14.2 million raised by Avaamo, which is building an AI-based chat service for enterprises
* $56.5 million raised by SparkCognition, a company that provides cybersecurity and machine learning application to businesses
While these are just the tip of the iceberg of AI startups around the world, it’s worth noting that none of them is focused 100% on digital marketing. Does it mean that marketers are safe, and don’t need to worry about AI taking away their job?
Maybe. PWC analysts predict that in the coming years, the biggest AI disruptors in digital marketing will be:
* Customised content creation
* Personalized marketing and advertising
Right now AI is doing great at recommending content to users based on what they’ve seen before. But that’s just the first step, and analysts suggest that down the way we will see this technology being used for automated telemarketing that feels like a real conversation, and in the even longer term – use-case specific and individualized AI-created content.
So AI will be able to talk to (and understand) our customers and create super-personalized content for them based on huge amounts of data that would take years for humans to analyze. It’s starting to look a bit bleak for us marketers, but those are just predictions. Instead of trying to paint a picture of what might yet come, let’s see how advanced AI is at the moment. Can it already perform fundamental tasks of digital marketers?
Writing constitutes a large part of digital marketing. Marketers write copy for paid advertisements, blog posts, cold emails, scenarios for video ads, research papers and more.
So far, AI hasn’t beaten us in creative writing. A lot of brain power goes into the complicated operations that result in a great piece of writing.
Not for lack of trying – new research is paving the way towards creating AI that will finally understand language. Currently, the most widely tested model is called Elmo – Embeddings from Language Models, released just this spring by the Allen Institute.
It goes to show how dramatically fast AI technology is improving. The Elmo model was better than previous comparable models at tasks like reading comprehension by as much as 25%.
But it’s not perfect yet. The main trouble in teaching AI to understand language is insufficient amounts of labelled data. We use language in unpredictable ways. Because of that, collecting enough data for AI to learn all about how we talk and write is virtually impossible. So the trick is to create models that can distinguish patterns from messy, unlabeled data.
It will take a lot of experimentation and a huge amount of brainpower to solve all challenges related to building AI that understands human language, and uses it like we do.
Marketers also have to create other types of content – images, audio, video.
A big breakthrough for image recognition came in 2012, when computers started recognizing images thanks to feeding millions of labelled images, from a database called ImageNet, into neural networks. This has made it possible to create self-driving cars, as well as build the robot that automatically tags people in your Facebook photos.
Can AI already create images for our blog posts and ads? Kind of, but not really. A good example of this is Google’s DeepDream – a program that finds and enhances patterns in images. It has been applied in tools that can combine two images into one, for instance, make your photo look like a Van Gogh painting. It can also do this:
In this example, the DeepDream neural network was taught to recognize dogs and run on an image of jellyfish. The results, as you can see, are pretty crazy, but far from the images that marketers create on a daily basis.
Another tool worth mentioning is Panels by Getty Images – a tool that uses AI to choose the right images for news stories. It doesn’t create the images, just picks them.
Can AI create music for our videos? Yes. There is a variety of open-source tools that allow you to do this, but those usually require a computer science degree and a certain level of genius to make them work. However, there already are products based on AI for music creation, which do the job perfectly – one of them is called Amper, and it’s childishly easy to use. You just specify what kind of music you want, and it generates the music for you.
Can AI create videos? Also yes. Perhaps not cinematographic masterpieces, but companies like Wibbitz or Wochit use AI to empower media companies and small creators to make engaging short-form videos optimized to engage viewers, and provide a database of material to use in their creations.
Managing digital marketing campaigns essentially requires collecting and analyzing large amounts of data. This is something that AI can do pretty well, and there are a lot of AI-powered tools already available for marketers in this area.
Artificial Intelligence is good at:
* Programmatic media buying
* Propensity modelling
* Predictiveness analysis
* Lead scoring
* Ad targeting
* Dynamic Pricing
* Web and App personalization
* Predictive customer service
* Marketing automation
But managing marketing campaigns with AI requires combining multiple tools, a well-thought-out strategy, and a deep understanding of the business and of target customers.
So AI is already helping us run campaigns, and our results often depend on how well we’re able to use algorithms to our advantage. The most common example here is the field of Search Engine Optimization – SEO optimizes content to make it more visible to search engines’ inherent algorithms.
Will AI make digital marketers obsolete?
Not likely. Even if AI learns to create all types of content, and run different digital campaigns by itself, it will still need a person to tell it what to do.
At the moment, marketers don’t have much to worry about. Most proponents of AI say that the aim of this technology isn’t to make human jobs disappear but to empower humans to generate better results with less work, and focus on high-level creative tasks while AI does all the repeatable, time-consuming tasks.
So AI won’t take our jobs, but its role in digital marketing is a very important one. Which means that we need to learn about this technology, and be on the lookout for new AI-powered tools that help us do our job better.
In the end, we must not be afraid of AI. We need to embrace it.