Marketers, meet the AI models that are poised to transform your industry.
Excited chatter about artificial intelligence has been sweeping across adland in recent months. “Generative AI,” in particular, seems to have already become the hot new marketing buzzphrase of 2023, similar to the hype surrounding “the metaverse” this time last year.
And like the metaverse, generative AI is technically complex, easily misunderstood and liable to a broad array of definitions and interpretations. Some starry-eyed advocates hail it as a harbinger of a glorious AI-assisted future; others shake their heads in dismay, regarding it as a pale imitation of human creativity – or worse. (Most of us have seen enough dystopian sci-fi movies to reflexively fear AI.)
But there is one foundational point on which almost all expert voices agree: AI is here to stay, and it’s only going to improve.
The ostrich’s strategy of sticking one’s head in the sand and hoping for a trend to disappear is an old parable of folly – one that seems especially well-suited to the rapidly evolving field of AI. The time has come for brands to get acquainted with intelligent machines, so that they’ll be able to face the future head-on.
What is generative AI?
Generative AI can be defined in a number of ways. Here’s a simple overview: Generative AI is a branch of machine learning which leverages algorithms, both unsupervised and semi-supervised, to create text, images, video, code and other content based on preexisting content.
Okay, deep breath. That’s a lot of new lingo to throw at you in one paragraph. Don’t worry too much about the details – if you ever want to dive into a technical rabbit hole, we recommend checking out this AI glossary which we’ll be updating regularly – for now, it’s enough to understand that generative AI models basically use old content to create new content.
ChatGPT is a generative AI model that’s received a substantial amount of media attention in recent weeks. Launched in November 2022 by San Francisco-based startup OpenAI, this model uses GPT-3 – a collection of large language models (LLMs) also developed by OpenAI – and text-based prompts to generate new text modeled after human conversation. ChatGPT might be new, but it’s already exhibiting some impressive abilities, including attempts at humor, as was recently demonstrated in an ad from Ryan Reynolds’ Mint Mobile brand.
Lensa AI is another generative AI model that’s recently been making waves across the zeitgeist. Users simply need to submit some selfies, and the program will create a set of digitally rendered images which range from photorealistic to abstract. You’ve probably seen some Lensa AI portraits cycling through your Instagram and/or Twitter feed starting around December 2022. Small wonder why: Most of those portraits are extremely flattering.
Then there’s DALL·E 2 and MidJourney, two other much-publicized generative AI models that use text prompts to create images. Both are free to use (MidJourney requires a paid subscription after the first 25 images) and are currently in beta.
As excitement about the capabilities of generative AI has soared, so too have some ethical concerns. The case has been made, for example, that AI models like Midjourney and Lensa AI are nonconsentually using the artistic styles of human artists. Meanwhile, some claim that ChatGPT could spread dangerous misinformation, propagate harmful biases and put to a variety of other nefarious uses.
‘This is not hype’
It takes a lot of data to keep generative AI models up and running.
The human brain is ill-equipped to comprehend such enormous numbers. GPT-3, for instance, uses around 175bn parameters in order to predict the next word that should be added onto a sentence. (That’s 175 followed by nine zeros.) Google’s LaMDA, another LLM-based generative AI model, was originally created using a dataset of approximately 1.56tn words (12 zeros) according to a company blog post. LaMDA would prove to be so adept at mimicking human dialogue that one of the company’s former engineers felt obligated to publicly speak out, in July of last year, claiming that the program had become sentient. (Google fired the engineer and denies those claims.)
The scope and scale of these models underscores the human brain’s sophisticated, dynamic and often mysterious language-processing abilities. “Language is a complex thing – it’s a very high-level intuition that humans have,” says Rajkumar Venkatesan, a professor at the University of Virginia Darden School of Business whose research focuses in part on the impact of AI on the marketing industry. “It’s a big breakthrough to have a machine that can converse [and] understand the nuance of language.”
Venkatesan points specifically to the recent Mint Mobile as evidence of the breakthrough he’s describing: “Humor is a highly intuitive thing to pull off … and [ChatGPT] actually cracking a joke? That is some level of intelligence for a machine.” He adds that some of the more advanced models of generative AI also seem able to understand the more subtle aspects of human conversation, like sarcasm and double negatives.
“This is not hype,” he says. “It’s only going to get better as we keep going and giving it information.”
Should marketers be worried?
It’d probably be a good idea at this point to pause and take another deep breath.
All this talk of trillions of data points and the rise of joke-telling machines raises some disconcerting questions about the future of humanity; these have been discussed at great length elsewhere and, again, Hollywood has for decades been making films exploring the dark side of AI, so we’re not going to rehash all of that here.
Suffice it to say that concerns about intelligent machines tend to fall into one of two basic categories: (A) they’re eventually going to become smarter than every living human being (a hypothetical event which computer scientist Ray Kurzweil has dubbed “the Singularity”), after which point they will be completely and irrevocably outside of our control, or (B) even if they don’t become our intellectual superiors, they will become sophisticated enough to push huge numbers of people out of the job market. (Some believe that that second outcome might not be such a bad thing, because those people will be able to enjoy more leisure time, but that’s a whole separate issue.)
Many marketers are probably looking around at all of the creative work that’s being generated by programs like ChatGPT and MidJourney and are asking themselves: Will there be a role for me to play in five years? Ten? Twenty?
“That’s a real worry,” says Venkatesan. “There will be a shift in the creative side … anywhere there’s a restructuring in the economy and the labor force has to learn a new technology, there is going to be pain.” Meaning, brands are likely to restructure their creative strategies as the means of producing creative content continue to change in the coming years.
At the same time, Venkatesan also believes that the rise of generative AI could very well increase the demand for highly talented human employees – those whose abilities rise above those of a machine like ChatGPT, which Venkatesan says can write “a ‘B’ paper, not an ‘A’ paper.” Generative AI “is going to make some good copywriters really valuable,” he says. “Anybody who’s above-average, the premium they can command is going to be huge. That’s the bright side.”
Just as brands have had to hire social media copywriters and managers over the past ten-plus years in order to thrive in that crucial new media, brands may also need to hire generative AI specialists in the coming years. “You will need folks who are good at using it and making it do things you want it to,” says Venkatesan. “That’s a new job that I think is going to come up.” One can almost imagine the LinkedIn job posting now: “ChatGPT prompt expert wanted.”
The bottom line? “There’s going to be disruption in the creative advertising industry, for sure,” says Venkatesan.
Deloitte chief marketing officer Scott Mager agrees that there will be some disruption, but he’s generally optimistic about the AI-generated future of the marketing industry. “AI tools might replace a designer that doesn’t know generative AI tools, but I don’t think it’s going to replace designers or copywriters,” he says. “It’s a creative renaissance; it puts creativity in more people’s hands, and I think we’re going to get more interesting content that’s [currently] locked in people’s heads.”
For more on the latest happening in tech, sign up for The Drum’s Inside the Metaverse weekly newsletter here.