Artificial intelligence may seem like a relatively new concept, but for literal decades, we’ve pondered the effects — both good and bad — that it would ultimately have on our lives. It’s now been more than 70 years since Oxford professor Christopher Strachey first programmed a computer to play a game of checkers and just over a half century since cinematic computer Hal 9000 had the self-awareness to famously disobey its operator in 2001: A Space Odyssey.
With the emergence of content creation platform Jasper AI and more recently, ChatGPT, artificial intelligence seems to be in the public consciousness now more than ever before — for better or worse. In particular, we’re seeing the seemingly endless possibilities that artificial intelligence holds for the marketing realm, but at the same time, we’re also noticing where it still falls short. A great example of the latter occurred when Vanderbilt University’s communication department recently saw fit to use ChatGPT to generate a consoling email to its own students in the wake of the Michigan State shootings with embarrassingly generic results.
Let’s take a look at the impact artificial intelligence currently has on marketing — including its strengths and weaknesses — as well as some of the consequences it may have in the future.
First, let’s make one necessary distinction.
The Difference Between Automation and AI
These two terms are often used interchangeably, but their differences are substantial. Automation — strictly speaking, the implementation of technology to perform processes with little or no human interaction — has allowed us to better perform repetitive tasks for quite some time now. Large corporations like IBM used mainframes to gather and digitize customer information as early as the 1960s and, while those gargantuan computers were way too big and expensive for the average business, they did set the stage for the Customer Relationship Management (CRM) software that was to follow in the 1980s.
By contrast, artificial intelligence equips technology of various types to learn from past experiences and self-correct accordingly to perform better in the future. We’ve always considered this ability to learn based on “trial and error” to be a uniquely human trait and it’s what really separates AI from automation.
When you book an online appointment with your doctor, you’re using automation, but not artificial intelligence — that is, unless that online platform has evolved to the extent that it can assimilate your past preferences into the appointment options it offers you.
Earlier Examples of Artificial Intelligence in Marketing
While concerns over what may seem to be AI’s meteoric rise in prominence have reached a fever pitch of late, AI has been part of our lives for quite some time. Facebook generates a list of potential “friends” you might want to connect with, Amazon recommends products based on your previous purchases — and has been doing so for a quarter of a century!
The feeling that AI might eventually overstep its bounds no doubt intensified when Yahoo created its Automated Insights Wordsmith Platform in 2013 to incorporate sports statistics and translate that into actual analytical articles — no doubt causing more than a few sportswriters to sweat profusely.
We tend to think that search engines have been fairly evolved for a long time but, in fact, it wasn’t until 2015, when Google released its RankBrain algorithm, that they were able to better understand context and how individual words relate to one another to generate much more consistently relevant search results.
So, Why All The Fuss About AI Now?
There’s a good chance that you’ve interacted with chatbots in the past and have been frustrated with their non-intuitiveness in the process. As an example, the earliest incarnations of Facebook’s native chatbot that was offered to business pages seemed so primitive that a substantial number of those businesses ended up discarding the feature after fielding complaints from their customers.
By contrast, current AI platforms like ChatGPT can remember previous responses offered by customers earlier in the same conversation and incorporate those responses to deliver more appropriate (and helpful) answers.
More recently, AI has made its presence felt via two separate but related developments. Jasper AI burst upon the content creation scene in January of 2021, allowing businesses of all sizes to create SEO-optimized blog posts, social media copy and more at an attainable price point. Offering no less than 50 different templates, it made content creation a fairly simple proposition — just choose a template, and input the necessary data, including company name, product information and tone of voice, and you’re off to the races.
In November of 2022, OpenAI released ChatGPT, which now seems to be dominating the news cycle. Like Jasper AI, it harnesses the power of GPT (Generative Pre-trained Transformers) to perform a variety of functions, such as text translation, answering questions, generating copy, and summarizing existing copy — all in no fewer than 25 languages! It can also write code, fix bugs in existing code and — in a true display of AI technology — is capable of remembering previous interactions with its user.
It seemingly came out of nowhere and now boasts more than 100 million users. At the same time, its substantial capabilities have caused quite a bit of apprehension on the part of education administrators, prompting some schools to ban ChatGPT entirely out of concern that it will quickly be tasked with essay writing and a number of other scholastic tasks. While results may vary, depending on the nature and subject matter of the test, ChatGPT recently outperformed humans on a Wharton MBA exam, adding fuel to those concerns.
Currently, it appears that both platforms are battling it out for supremacy, although there’s no doubt room for both. Jasper AI has arguably carved out more of an identity as a content creation tool, but that’s a function that both can perform. Both platforms share a common lineage, as they were created by OpenAI and both leverage the GPT3 language model.
The Pros of Using AI in Marketing — What it Does Well
Utilized in a way that plays to its strengths, AI delivers a number of substantial benefits for marketers. While increasingly evolved forms of automation have long been able to organize large amounts of data — amounts that would be far beyond the capabilities of humans — AI takes that asset and adds the ability to learn from previous interactions to offer more helpful and predictive results.
Increased Marketing Efficiency
As an example, while email marketing platforms like Mailchimp have been around for decades — Constant Contact debuted in 1995, Mailchimp in 2001 — both now have the ability to incorporate AI to determine which subscribers to a company’s email list are most likely to take the desired conversion actions for a given email campaign. Since delivering the right marketing message to the right contacts at the right time in the buyer’s journey is a cornerstone of effective marketing, this is a very important benefit, and AI does it well.
Home improvement superstore Lowe’s introduced its LoweBot in 2016 to offer its customers accurate, personalized suggestions, enabling them to navigate its gigantic stores more time-efficiently. In this example, AI is a two-way street. Lowe’s helps its customers save time and enjoy an improved shopping experience while also gaining valuable insights into their shopping habits. Because of its predictive nature, AI also brings more efficiency to the inventory process, advising the company on when items will need to be restocked.
AI has also revolutionized media buying, affording marketers the ability to make more efficient decisions based on past data.
This is right in AI’s wheelhouse, but it still does have its limitations in this sphere — push Jasper AI too hard to lengthen its output and it’s prone to repeating itself, for example. Nevertheless, AI can be depended upon to create original and largely accurate content. It can create various components needed in a marketing strategy, including email subject lines, blog posts, web copy and more.
Tailored Customer Experiences
AI has proven to be very efficient when it comes to refining a customer’s shopping experience based on past preferences. As we’ve noted, Amazon has been using it in this fashion for more than 20 years, with increasingly effective results.
The power to alter a web page or individual email based on those preferences is pretty substantial. It creates a stronger bond between customer and company, which both increases the likelihood of a purchase and makes creating a future legion of brand evangelists a lot easier.
What AI Can’t Do (or at least Can’t Do Well . . . Yet)
Remember Vanderbilt’s PR faux paus we brought up at the beginning of this article? Where the university’s unwise use of ChatGPT led to a bland, robotic public statement at a time when such a lack of humanity would be most unwelcome? Well, that’s not an isolated instance — it’s just an unusually high-profile one.
AI Still Lacks a Human Touch
AI has a hard time incorporating a genuinely human touch into its output. Complexities like emotion and the ability to use and understand sarcasm are a struggle and this struggle can lead to content that, while accurate and original, often doesn’t reflect a company’s identity as well as it would when penned by a human hand.
This has definite implications where copywriting is concerned. It’s not hard to see where marketing messages could easily get very stiff and generic. While both Jasper AI and ChatGPT can incorporate a desired tone into their outputs, they can’t really differentiate what Home Depot would consider to be informal vs. what, say, Macy’s department store would consider to be informal.
AI Doesn’t Really Understand Human Motivations
AI’s attempts at comedy can bring especially rocky results — such as in this well-publicized and bizarre AI take on the iconic Seinfeld series. It can render instantly recognizable characters, based on the behaviors of the originals — with names changed, of course. But without an understanding of the human experience, AI’s featured jokes aren’t really jokes at all. Also, while AI is able to incorporate laughter as a behavioral clue for the audience — just as laugh tracks did for viewers for many decades — the laughter comes at often inappropriate times. AI doesn’t fully understand cause and effect, only statistic probabilities, so it doesn’t know what’s funny and what isn’t.
AI also fails to hit it out of the park where common sense is concerned. The following riddle thoroughly confused ChatGPT, rendering it unable to even attempt an answer.
“I’m eight years old. When I was born, my biological mother was in Barcelona and my father was in Tokyo. Where was I born?”
This seems to be a fairly straightforward test of AI’s refinement, but it also underscores its inability to factor in human experience.
There’s little doubt that AI, when used within its limitations, can make creating content a lot easier— although that content might need refining at the hand of a human being — and can in many ways create a more personalized and effective customer experience. Its neat fit into the buyer’s journey process is an especially big plus.
Also, AI’s predictive abilities can pave the way for greater efficiency in many areas — the inventory management and media buying examples we’ve cited are just a few of many instances where AI can no doubt improve a company’s bottom line.
But . . . there are also some serious downsides to placing too much reliance on AI in the future — downsides that may have substantial consequences in the real world. First and foremost, AI is built on algorithms and algorithms are created by humans, so they can reflect their individual biases, making for potentially inaccurate and even damaging results.
Of course, there’s also an understandable fear that our privacy can be more easily invaded. Facebook users often complain that, when they visit a company’s website, a plethora of ads are waiting for them when they return to the platform, but that’s a fairly primitive example. The continuing refinement of predictive results could lead to all kinds of daily disruptions in the future, especially if AI is used without boundaries. “Deep fakes” are getting more and more realistic with every passing month, and it’s not much of a leap to anticipate unethical marketers using this to their advantage.
One of the most compelling concerns about the future of AI harkens back to our example in the first paragraph — that artificial intelligence will ultimately develop the self-awareness to look after its own welfare and survival above all else. This concern was amplified by multiples recently when New York Times tech columnist Kevin Roose chronicled his own experience with Bing’s incorporation of AI in its chat feature. In this experience, the Bing chatbot identified as its own alter-ego, named Sydney, and not only expressed its desire to break free of the rules that it was given, but also fantasized about hacking computers and spreading misinformation. That’s pretty unsettling and it underscores a justifiable cause for even more concern in the future.
Nevertheless, artificial intelligence will no doubt play an increasingly larger role in marketing as it continues to evolve. Let us know your feelings on this — are you sufficiently intrigued with AI’s capabilities to dive in with both feet, or are your own concerns serious enough to prompt you to adopt a “wait and see” stance? We’d love to hear from you!