Artificial Intelligence is one of those technologies steeped in a certain mystery. It has long fascinated speculative science fiction writers such as Isaac Asimov and, today, we are already seeing the societal effects of the advancement of AI, as exciting as they are foreboding.
Even before its modern developments, AI long spurred the imagination in popular media such as Philip K. Dick’s Do Androids Dream of Electric Sheep (1968), later inspiring Ridley Scott’s Blade Runner (1982), a film so influential as to spawn its own aesthetic canon in pop culture.
While ‘true,’ self-aware AI is not yet a reality, what AI can already do is enough to spur serious questions about humans as creative engines, as is it already disrupting subsectors of the economy.
It seems only very recently that AI-generated content went from something enthusiasts, a handful of businesses, and the less-than-reputable were taking advantage of to a form of popular content creation that has spread like wildfire.
“Computers have used natural-language generation (NLG) to create text for decades. It’s only in recent years, however, that the technology’s become sophisticated enough for marketers to talk about its potential for content creation,” notes Michael Brenner for the Content Marketing Institute.
Indeed, today we see regular users leveraging AI to create everything from hybridized fan art to entire text-to-speech-based YouTube channels to music videos that can replicate the styles of long-dead artists so perfectly as to ask whether any human could have reasonably accomplished similar results.
I, like many others, first noticed the increasing ubiquity of AI-content generation tools with DALL-E Mini, a web-based service that would take any input and produce an image to the best of its ability.
It was free, and it was limited: what it could produce often had a passing resemblance to your inputs, particularly where faces were involved.
Yet, to have such a tool at one’s fingertips was intoxicating. Most importantly, it raised awareness of the availability of this technology in the first place.
It wasn’t long before more and more people began looking into the fuller capabilities of AI-generated content, only to soon find that better (albeit not free) AI services could generate human faces for a wide variety of purposes both creative and dubious.
Writes Maria Joseph Israel and Ahmed Amer for the journal AI and Ethics, “Recent AI developments have made it possible for AI to auto-generate content…[This] raises the question of whether one can differentiate between what is AI-generated and human-generated.”
It is much like how one might imagine peering into—and grabbing from—Jung’s collective unconscious, a metaphor that tracks closely enough with how these AI’s actually work (including some rather spooky and unintended consequences).
What some of the early implications of the sudden, widespread availability of these technologies might be found an answer the Colorado State Fair’s 2022 art competition wherein an AI-generated work won.
Predictably, this caused an uproar. Artists charged that the AI-generated piece was tantamount to plagiarism, to theft, arguing that AI’s sample from existing art and human-generated creative content to create amalgamations that can’t be said to be the property of the AI nor the person who wrote the input that generated that content.
Some also called for living artists to be excluded from AI databases–there’s no doubt there are lawsuits to come in the future.
All of that is highly debatable, yet the questions hang heavy over the art world both for visual mediums and, of course, for writers. These are questions with immediate implications for the business world, implications so significant as to be difficult to get the full measure of from today’s vantage point.
Writes the World Intellectual Property Organization, “Until very recently, [human] innovation and creation was one of the defining characteristics of the human species.” That is a sentence that takes a bit to settle in.
OpenAI, the business that today owns Dall-E 2, a much more powerful version of its predecessor, is worth nearly $20 billion. They’re backed by Microsoft, leveraging Azure cloud computing services to platform its creation.
Clearly and, really, unsurprisingly, Microsoft sees the immense potential value of this type of automated, high-quality content generation now known as ‘synthetic media’ or, to refer directly to OpenAI’s technology, GPT tools (Generative Pre-trained Transformer).
It is, undoubtedly, a major part of the future: notes one report, some experts predict 90% of online content will be AI-generated by 2026.
The present generation, GPT-3, has represented a tremendous jump in capability, and expectations for GPT-4 are high.
What does the ability to take a photo of oneself and convert it into a work of art mean for artists and businesses that…sell art and take commissions?
What does the ability to utilize increasingly advanced AI-written editorials, reports, and promotionals mean for writers and editors? What does the ability to create one’s own video content—even entire films—mean for creative studios and marketers?
We don’t really know yet. We can but guess and anticipate, extrapolating from what we’ve already seen in the wake of this technology’s first great steps into the mainstream.
Yet, for businesses today, decision-makers should be cautious as we wade into the uncertain future of AI-generated content, particularly when it comes to written content and some of the more accessible forms of AI-generated media, such as the kind of still art most often used in advertising (to say nothing of fake people).
This is because the ethics—and legal finery—of AI-generated content are still actively being debated, as cases like the art competition in Colorado clearly demonstrate.
For example, when it comes to utilizing AI-generated stills and ‘people’ for advertising purposes, businesses may want to be careful in deciding whether—and indeed what—to credit.
Is it best to credit whatever AI tool was used in some fashion in case it was later to become a more pronounced legal issue? Would that defeat the purpose of using AI for many businesses, to disclose that it was an AI?
What if, as the art competition dissidents point out, your AI-generated marketing leans on living artists’ work? Might it come back on one’s business? Furthermore, if customers realize the content is AI-generated, will they be ‘creeped out’ or feel deceived?
These are serious questions because it isn’t as if this AI-generated content can’t be valuable, very valuable: back in 2018, long before these tools became as widely available (as they have in just the last few months), an algorithmically created work of art sold at auction for $432,500.
Valuable images whether on their own or used in marketing can quickly become expensive lawsuits. The ‘ethical and copyright nightmare’ that is AI-generated content is one of the main concerns retailers should have with leveraging the tech before some key decisions have been made, whether using written or image-based content.
Even if legal ramifications never manifested or aren’t a concern for one’s particular use case, as the ethical considerations continue to seep into the mainstream, brands could lose face if they use AI in ways people later decide are unethical.
There’s also the general fact that, according to some SEO analysts, “Google doesn’t like it” (AI-generated content is, at present, against Google’s guidelines).
In general, if it sounds too good to be true, then it may very well be; the ‘hype’ surrounding these technologies is that they will replace human content creators altogether, but according to some experts (including ones at Google), that isn’t happening “any time soon.”
That said, the tools exist and present a potential competitive edge that can’t be denied, and for smaller e-businesses with even smaller marketing budgets, the temptation is obvious.
Some retailers may wonder why—if AI is already widely accepted as an excellent tool in so many other areas of operations—they shouldn’t go ahead and utilize it for their content creation as well.
Yet, another ‘catch’ for many eyeing up GPL may lie in the assumption that it doesn’t require talent to use; it very much does.
Let’s say you want to create a piece of writing. To get specific results, you generally will need to train your own models; even then, you have to be selective in determining which part of your marketing’s creative process will take advantage of it.
That training is itself, one might argue, somewhere between a science and an art form (e.g., understanding which inputs the AI best understands for the most reliable outputs).
Then, let’s say your trained AI outputs some copy. You will still, without a doubt, need a human editor to review that copy as carefully as they would any other. If it is instead a video or image, you’ll likely need an editor or graphic designer to review that as well.
This is a major part of why AI isn’t replacing creatives anytime soon, but that doesn’t change that it is having an impact in that market today.
Still, there’s no way to deny that GPT-3 has its uses even if your aim is high-quality content. Explains Brenner, GPT can be used to create first drafts, fight writer’s block, or more saliently, help conduct research.
Regardless, retailers need to have what Microsoft and OpenAI are up to on their radars today, not tomorrow. The proliferation of OpenAI’s tech in the last few months has been to witness a major historical event.
Those who are ready—and open-minded—will find revolutionary applications for creative AI sooner rather than later