Beyond the Hype Train: The Myths and Realities of AI in Marketing
While the term has been around for a while, the technology, at its core, is still very much in its nascent stages.
We are excited to announce the first round of leaders who will bring our 2020 theme HUMAN.X to life at the Broad stage this June (17-18).
There is no doubt that among the vast developments in business processes, Artificial Intelligence has emerged as one of the hottest buzzwords. This is especially true with regard to the continually advancing effects of AI in marketing.
However, despite the growing hype and evidence based on case studies and statistics, not everyone is jumping on the AI train—all because of various myths and misconceptions surrounding this revolutionary innovation.
While the term has been around for a while, the technology, at its core, is still very much in its nascent stages. As such, there remains some confusion as to what constitutes AI, its practical applications, and what it means for people and organizations.
This article touches on the myths and realities surrounding the present state of AI.
First off, what exactly do we mean by Artificial Intelligence?
This is an important starting point in any discussion of Artificial Intelligence.
AI is a broad term that could mean a plethora of things—a fact that has led to the massive hype behind it. In its current forms, AI has become an umbrella term that pertains to different technologies that include machine learning, deep learning, and neural networks.
As noted by McKinsey Global Institute partner Michael Chui during an episode of the McKinsey Podcast, AI was a term first used half a century ago by Alan Turing. He explains that, currently, it refers to using machines to do things considered to be “intelligent” – or being able to either simulate or do things often prescribed to things people do with their cognitive faculties.
The Marketing Artificial Intelligence Institute points out that it’s important to keep in mind that some AI commentators treat the different technological concepts under AI as interchangeable. They are not. Instead, AI has become a term referring to a suite of tools.
Some of the simpler applications of these technologies include:
- Physical AI – robotics, and self-driving cars
- Computer Vision – image and video processing
- Natural Language Processing – chatbots, predictive texting
For now, at least, AI has become a generalized marketing term. AI is not a magical, singular solution–at least not yet Understanding this is crucial in helping you begin to separate the hype from the realities of AI.
But there are now a plethora of tools that can help with tasks – that while may seem rudimentary – can be surprisingly helpful to drive business growth. Find out what areas of your business need improvement, and see if there’s an innovative technology that can drive the desired business result.
Here are a few other myths to keep in mind when it comes to AI in Marketing:
Myth #1: AI will replace human jobs
No, at least not in the context of doomsday scenarios.
Certainly, there will be industries (like autonomous trucking, and, at least in scale, customer service) that will be affected by the continued evolution of AI. But foreseeably, it will be professionals whose jobs (and skill set) don’t evolve that will be in jeopardy.
But it doesn’t mean that only low-skilled and manual workers will be replaced by AI.
In the same way, AI-equipped robots and machines have already taken tasks off of professionals, such as doctors and lawyers. Machines help lawyers by scanning mountains of documents at lightning speed. In the medical field, machine learning algorithms are able to assess scans and x-rays, searching for early warning signs of diseases.
Similar to how marketing automation didn’t mean the extinction of marketing professionals, the continued rise of AI should see people’s work enhanced by the technology.
As stated by Tomasz Zietek of Fornax.AI in an interview , professionals can focus on higher-value tasks with AI taking over certain rudimentary tasks. This ultimately enables them to become more productive, performance-driven, and afford them more time.
Just as consumers, technology, and the way we consume media have evolved – so will the tasks entailed by jobs. If you have a job that predominantly consists of tasks that AI can automate, you have to adapt and evolve – or otherwise can see yourself become dispensable.
Yes, AI will be removing certain roles from humans – but they’ll also be creating entirely new ones.
Myth # 2: Adding AI to the marketing mix is too expensive
This may be true for small to mid-sized organizations who are, currently, still a ways away from being truly AI-ready. Mid-sized organizations that don’t have an established and structured data collection process may also find it beyond their means to add AI-related systems to its marketing mix.
Of course, innovation does not come cheap. In fact, for AI marketing companies to start operations, a number of them relied on multiple rounds of seed funding that amounts to millions of dollars.
Thanks to those investments, however, AI tools are now more accessible.
An increasing number of companies can now benefit greatly from existing AI marketing tools like chatbots, and image processing and captioning that can make business processes that much more efficient. The key here is knowing how to apply the technology in a way that would positively affect the business’ bottom line.
If you are able to pinpoint which aspects of your business process can be aided by existing tools that fall under the AI umbrella, then you won’t need a budget that only companies like Apple can come up with. As pointed out by Information Science CEO, Seth Earley , not everyone needs a team of data scientists and machine learning experts to adapt AI tools.
A huge appeal of AI tools is its low cost but high levels of efficiency. For example, there are a number of tools that can help with various marketing aspects like web design, content creation, or advertising.
For instance, The Grid –the creator behind the AI “Molly”–can design a website after you input your content (e.g. copy, CTAs), effectively taking away the need for a full team of developers and software engineers. The best part is that it starts at $100 per year, which is a lot less expensive than full-time salaries.
There are also tools that will help content creators make their tasks more efficient. Forbes and The Associated Press already use tools like Wordsmith and Quill –both platforms that can turn data into human-sounding narratives. Click here for a sample of an automated content.
You can also use a free tool called The Hemingway App to check the grammar of your articles.
Myth #3: AI algorithm can make sense of any data
While data is the most important input for existing AI tools, it doesn’t mean that any data will do. If you’re looking to adapt AI tools for your business, it’s important to first establish that you have the right, pointed data to feed into the system.
As noted by Earley, it is erroneous to assume that you can just point an AI system at any data set and expect its algorithm to come up with the desired answers and solutions. With the vastly improved data collections processes and the computing power of machines, data that is either unprocessed or too broad can negatively impact the performance of AI tools.
For an AI tool to reach optimal performance, a carefully curated, high-quality data set is necessary. Keep in mind that AI is not the magical machine portrayed by Hollywood movies; AI is, in reality, composed of mathematics, patterns, and iterations.
For you to be able to successfully adapt, you need to first understand its three core components :
- Training Data (TD) — the initial data that the AI tool will learn. TD has inputs and pre-answered outputs, which the machine can then use to detect patterns (E.g. input is a customer service email thread with a CSR; output is the business’ support ticket categorization level).
- Machine Learning (ML) — this is where the machine learns the pattern and applies that to subsequent inputs (E.g. when a new customer service email thread is received, the machine would then predict a categorization and include its level of confidence for that prediction). ML learns as it goes. Instead of sticking with fixed rules, it adjusts as the more it processes additional TD.
- Human-in-the-loop (HITL) — this is where humans are needed. AI is not foolproof; you need an expert to check the machine’s output to make sure it’s 100% accurate.
You would, therefore, need to “train” your AI tools and configure these components according to the needs of your business. As the Information Science CEO wrote, an algorithm is a program, and programs need good data. If you don’t have the right data, then you’re not being efficient with your use of AI.
Myth #4: AI can’t control people
To outsiders looking into the tech world, and more specifically, the continually growing realm of AI – the human ego has a tendency to think that just because data scientists and programmers design the algorithms that AI systems are dependent on, that they are incapable of controlling people.
For quite a while now, digital marketers utilize AI systems to manipulate our thoughts, feelings, and emotions. How many times have you Googled a product/service and see that exact thing “magically” appearing on your social media feeds?
Companies like Facebook and Amazon keep track of everything you search for. Then, that information is used for what marketers call “retargeting”. That is, to continually show you the same products (through adverts, sponsored posts, etc.) to influence your interests – and ultimately, your purchasing decision.
Current AI systems may not yet be able to completely control humans, but for years now, they have certainly had a huge hand in influencing their consumer behavior.
Myth #5: Free-thinking AI isn’t coming
Yes and no. Marketers have been using machine learning to help come up with more effective and efficient solutions for a while now. But machine learning still largely depends on the continuous re-approximation of available data sets to come up with better solutions.
In the vast spectrum of AI, there is a corner known as “deep learning,” a machine learning technique that imitates the way humans’ biological neurons learn patterns. This would then allow a computer to mimic the complexity of human behavior.
Currently, deep learning is used for simpler tasks like voice recognition and language translation. But as pointed out by Technology Review , deep learning is also being developed to accomplish more advanced tasks like diagnosing fatal diseases or making billion-dollar trading decisions.
Chip manufacturing company Nvidia, for example, developed an experimental autonomous car that’s unlike any of its Google or Tesla brethren. Instead of relying on an algorithm made up of instructions on how to drive itself, it followed an algorithm that taught itself how to drive by watching a human do so. What some experts find unsettling in this development, is that how the machine is making its decisions isn’t exactly clear.
This uncertainty is proof that free-thinking AI may not be too far ahead in the future as some might imagine.
- Practical applications of AI
- Customer Retention
At the end of the day, the pursuit of driving revenue growth drives any process or tech adoption. And increasing customer retention is a key ingredient in pushing this.
As noted by marketing automation software company Emarsys, if a company can increase its retention rate even by just 5 percent, its profits can go up anywhere between 25 and 95 percent. And that doesn’t even include findings that state customer acquisition costs five times more than retaining existing customers.
An infographic on RedStag also shows the excellent effects of AI in engaging and retaining customers, particularly in ecommerce.
Some of the tactics employed to do so include the personalization of customer experience, and the presence of virtual buying assistants. Plus, conversational marketing using chatbots and voice-activated applications. Other methods include providing educational and other valuable content, and special offers with the use of email and social media.
With the vast ocean of data collected, it became harder and harder to accurately target the right audience. AI-powered segmentation engines allows marketers to focus beyond data collection and into high-value aspects such as:
- Understanding individual buyer’s purchasing behavior
- Automating segmentation and communication
- Creating effective, highly personalized interactions
Study shows that customers use an average of 4.5 devices when using the internet . And with customer expectations elevated to previously unforeseen levels, brands are scrambling to provide an omnichannel experience to consumers.
AI technology allows brands to create a consistent, efficient, and seamless customer experience at every touch point. All the while taking care of rudimentary (albeit, time-consuming) tasks, and allowing marketers to focus on creative strategies.
Hollywood had a lot to do with inflating expectations for AI in marketing, way before any of the technologies under the AI umbrella even started their development phase. And fairly recently, real-life Tony Stark, Elon Musk has fanned the flames by stating that in the future, whoever controls superior AI will be the world superpower.
While not every company is ready to adopt AI-based tools today, structurally and philosophically preparing their organization is necessary to avoid lagging behind when Artificial Intelligence use becomes inevitable. However, you also can’t forget about acclimating your employees to the concept. You might have all the best AI-powered tools, but if your employees don’t fully embrace the concept, you won’t see the impact you desire in your business.
Learn the latest trends, insights and best practices from the brightest minds in media and technology. Sign up for SMW Insider to watch full-length sessions from official Social Media Week conferences live and on-demand.
Write for Us
Interested in sharing your ideas and insights with the world? Become a SMW News contributor and reach 300k readers each month.