The buzz around ChatGPT (Generative Pre-trained Transformer) took flight on February 7, 2023, when OpenAI — an artificial intelligence research and deployment company — asked a limited number of people to test it. The initial reviews were rave, and by the end of March millions of people worldwide were using ChatGPT to create content. A student in the UK asked ChatGPT to write a letter to the city council protesting a parking ticket, and her fine was revoked. And this is just the beginning.

For the uninitiated, if there are any of you left out there, ChatGPT is one of the most popular and widely used large language models (LLMs) in circulation. ChatGPT can follow complex instructions given in spoken language and solve challenging problems accurately — imagine if Alexa or Siri could contribute to a creative process. Its capabilities are as astonishing as they are simple:

  • generate, edit, and revise in collaboration with users on creative/technical writing tasks, including songwriting and screenwriting
  • learn an individual user’s writing style
  • accept images for inputs and generate captions, classifications, and analyses
  • handle more than 25K words and support longform content creation, extended conversations, and document search/analysis.

Google, Facebook, Microsoft, and other tech companies have developed similar models. And AI is stretching far beyond just smart chatbots. There are already several iterations of:

  • text-to-image models that create realistic images from natural language
  • image captioning models that describe pictures with words
  • open-source applications that will classify and summarize text
  • vision libraries that enable computers to detect and track objects.

These tools have spurred a tremendous amount of both excitement and distress in the world of work. Will AI take our jobs? Make us more productive? Or a combination of both?

Let’s examine ChatGPT as it’s used in the advertising industry specifically for a closer look.

The Limitations of AI

From OpenAI, the developer of ChatGPT: “While we have safeguards in place, the system may occasionally generate incorrect or misleading information and produce offensive or biased content. It is not intended to give advice.” OpenAI admits to three core limitations of ChatGPT:

  • may occasionally generate incorrect information
  • may occasionally produce harmful instructions or biased content
  • limited knowledge of world and events after 2021.

These limitations bespeak the need for human participation in any and every AI project. According to Todd Reinhart and Bill Skrief of AdAge, AI should streamline the human creative process, but it shouldn’t produce final deliverables:

“Just because a technology is in use doesn’t mean it’s ready for prime time. AI has been a media darling, but creative leaders need to look beyond the hype to determine if AI is right for their process and internal needs and if it synergizes with the comfort level and requirements of clients and partners.”

Reinhart and Skrief warn of the necessity to properly govern AI. Essentially, users must be fully aware and transparent about what it can and can’t do, as well as what it should and shouldn’t do.

For example, FreedomGPT is an LLM-based chatbot trained to have neither guardrails nor inhibitions. The tool was built by AI venture capital firm Age of AI and dispenses with most forms of censorship: it can explain how to build a bomb, it can use racial epithets and slurs and will, with the right input, happily praise Adolf Hitler. This has some unsettling implications. As recently as April, President Joe Biden told his council of science and technology advisors it remains to be seen if artificial intelligence is dangerous, but technology companies must ensure their products are safe.

Right now, ChatGPT is going through growing pains and working through a number of ethical and legal issues. It’s become apparent that human oversight is still required to recognize the technology’s potential for bias and toxicity.

How AI Can Help Ad Agencies

With only a few key phrases, ChatGPT can quickly generate multiple ad copy options. This can shorten the time usually devoted to brainstorming sessions or first drafts, giving copywriters and editors a head start. How simple is it? Ask ChatGPT, “write me ad copy for ______________.” So let it be written, so let it be done.

ChatGPT can also analyze consumer data and identify the most effective ways to reach specific consumer groups. The program can:

  • help advertisers determine the best social media platforms to use for their ad campaigns
  • determine the most effective times of day to target a particular market
  • generate new interests and audience groups.

ChatGPT can also conduct market research, another boon to ad agencies. This is immeasurably useful for advertisers looking to target consumer groups and/or specific demographics and discern their likes, their interests, and their overall buying personas. Some advertising experts even predict ChatGPT will eventually provide wireframes for the development and deployment of individual campaigns.

How AI Can’t Help Ad Agencies

AI is a good tool for generating new ideas and first drafts, but from there, think of it as a cake pulled from the oven: you still need a pâtissier to ice and slice it and put it in a display case (which said pâtissier has set at the perfect temperature and arranged alluringly, etc.). An application like ChatGPT can give you a good starting point, but it takes human creativity, experience, and imagination to create something wholly memorable.

“…at best, ChatGPT is a tool, a powerful one, that can save time and help creative talent focus on bigger thinking, but at worst, it’s a shortcut to mediocrity at scale,” says M.T. Fletcher of AdAge. “You don’t need a gizmo; you need an idea. You need an idea for a story, because stories are what people remember. Make your consumer the protagonist, your brand the catalyst, and tell a tale of transformation, redemption or discovery. From myths to memes, stories are what we share. And campaigns are never-ending stories. Which means creativity and non-linear thinking are still the most powerful tools in your arsenal.”

Example: AI in Action

As an example of ChatGPT’s capabilities and limitations, we asked it the question in the title of this article: “AI and Advertising: How Should Agencies Balance Smart Tech and Human Talent?” Here’s its verbatim answer:

“Artificial intelligence (AI) has transformed the advertising industry, enabling advertisers to target specific audiences, optimize ad campaigns, and measure results with greater accuracy. However, despite the benefits of AI, it is important for advertising agencies to strike a balance between smart tech and human talent.

 Here are some ways that agencies can balance AI and human talent in advertising:

  1. Identify areas where AI can add value: Agencies should identify areas where AI can add value, such as targeting specific audiences, analyzing data, and optimizing ad campaigns. This will enable agencies to leverage AI to improve their advertising efforts while also allowing human talent to focus on more strategic and creative tasks.
  2. Hire talent with the right skills: Agencies should hire talent with the right skills to work alongside AI technology. This includes individuals who have a deep understanding of AI, data analysis, and technology, as well as those with a strong creative skill set who can think outside the box.
  3. Encourage collaboration: To strike a balance between AI and human talent, agencies should encourage collaboration between the two groups. This will enable team members to share knowledge and expertise, and to leverage each other’s strengths.
  4. Emphasize the importance of creativity: While AI can help optimize ad campaigns, it cannot replace human creativity. Agencies should emphasize the importance of creativity in advertising and encourage their teams to think creatively and come up with innovative ideas that resonate with audiences.
  5. Use data to inform decisions: AI can help agencies collect and analyze data, but it is important to remember that data is only one piece of the puzzle. Agencies should use data to inform their decisions, but also rely on the insights and expertise of human talent to make informed decisions.

In summary, advertising agencies should strike a balance between AI and human talent by identifying areas where AI can add value, hiring talent with the right skills, encouraging collaboration, emphasizing the importance of creativity, and using data to inform decisions. By doing so, agencies can leverage the benefits of AI while also leveraging the unique strengths of human talent.”

As you can see, it’s a great start, but it’s largely just a start. The article is grammatically sharp with solid pacing, syntax, and tone, but the content overall is very general and offers few unique insights that truly educate the reader.

How the Right People Can Put AI to Work for Your Agency

AI content will often be “pretty good,” and “pretty good” is impressive when produced with such ease and speed. But anyone in advertising, communications or marketing knows that “pretty good” isn’t enough, especially in an increasingly incredible competitive market during economic headwinds.

So will the phrase “flesh-and-blood” become the “bricks-and-mortar” of the ‘20s, a way to describe increasingly irrelevant human capital as AI continues producing more and better content? In short, no, it shouldn’t.

Replacing your copywriters, editors, and proofreaders with large language models, or your animators, graphic designers, and illustrators with text-to-image AI applications, is a recipe for disappointment and failure. If you put together an orchestra without a conductor or set sail without a captain, you’re sure to find yourself face to face with disaster. In other words, you can’t just leave AI to AI.

AI is likely to change how your agency works and how your talent spends their time. It can grant you efficiency and increased speed by enabling you to produce more content in shorter periods of time. But let’s not get ahead of ourselves — AI is not likely to outright replace employees en masse anytime soon. At least, not for smart agencies and forward-thinking talent.

Because AI is so accessible and affordable, it’s helping to establish a baseline of competency that every agency can easily reach. To stay competitive, it’s more important than ever to have the very best creative talent. Brands will need the writers, editors, project managers, analysts, and more who can write careful prompts, fact-check and edit AI output, fine-tune content for specific customers, propose novel approaches to common client challenges, delight audiences with original ideas and, perhaps most importantly, stay keenly aware of AI’s quickly evolving abilities and pitfalls.

Jobseekers who stubbornly stick to old methods may soon find themselves becoming more and more replaceable. But the open-minded creatives who are eager to learn new tools, adapt their processes, and consistently challenge conventional thinking are going to be the most impactful contributors that a company can have. This is the type of talent that successful agencies will be seeking out, developing, learning from, and utilizing in the years ahead, and it’s the type of talent we work with at Creative Circle.

We are in an age of breakneck progress in artificial intelligence — the chatbots have given way to AI tools that can create impressive, highly detailed images. Is it time to worry or rejoice?  

Say hello to an emerging and fast-evolving genre of AI known as text-to-image generation.

It’s a fascinating new front in artificial intelligence, where anyone can generate hyper-realistic images from a written text description.

While most users have generated work that leans toward the weird and absurd, like the Mona Lisa painting a portrait of Da Vinci, many are also experimenting with possible commercial applications. As you might imagine, this tech has stirred up deep existential and ethical questions about art, creating, and more. Who is the artist behind the creations — AI or its human user? Can machines be creative? And perhaps most germane for creatives — Will this new technology make certain creative industry jobs go *poof*?

Just this past midsummer 2022, a select few people in and adjacent to the tech industry were granted access to these text-to-image AI tools during initial beta testing. The two most prominent are Dall-E —derived from the name of the surrealist artist Salvador Dali and Pixar’s lovable animated robot Wall-E —and Midjourney. Dall-E was launched last year by OpenAI, a nonprofit research lab founded by Sam Altman, Peter Thiel, and Elon Musk, among others. Midjourney entered open beta in mid-July 2022 and comes from a self-funded AI research lab founded by David Holz.

These AI text-to-image tools are simple to use — but rife with controversy.

You can open the doors to a dazzling cornucopia of visual creation in seconds with just a few words or simple phrases. Here’s how these AI tools work: Users type in a text prompt like “a frog on a united states quarter” or “chickens gathered to watch human wrestling,” for example, and the results are wild. These programs can translate text into award-winning art that has roiled the art and design community.

The New York Times recently published an article, “An A.I.-Generated Picture Won an Art Prize. Artists Aren’t Happy,” chronicling the brouhaha around the Colorado State Fair’s annual art competition awarding Jason M. Allen, of Pueblo West, CO, with the blue-ribbon prize in Digital Art for a piece he had created with Midjourney (he won $300).

via Jason Allen

Allen’s work, “Théâtre D’opéra Spatial,” won the fair’s contest for emerging digital artists, making it one of the first AI-generated pieces to win such a prize, provoking fierce criticism from artists who accused him of “cheating.” Allen defended his work, which in submission he had explicitly labeled “Jason M. Allen via Midjourney.” After winning, Allen posted a photo of his prize piece to the Midjourney Discord chat, which made its way to Twitter, where it ignited heated debate and backlash. Here are some excerpts from the online mêlée.

  • “This is so gross. I can see how AI art can be beneficial, but claiming you’re an artist by generating one? Absolutely not,” shared one Twitter user.
  • “We’re watching the death of artistry unfold right before our eyes,” another Twitter user wrote, who was quoted in the New York Times piece.
  • “No effort? Please,” another wrote. “If Jackson Pollock can splatter paint onto a canvas or Maurizio Cattelan can tape a banana to a wall, and both are called “art” (both which take hardly “any effort at all”), then this counts too.”
  • “Fine tuning and curating is the art here. If they just presented generic Midjourney art, then… It wouldn’t have won. Figuring out what looks like good digital art is the art itself.”

Some tweets excoriated Allen, while others defended him. Many argued that using AI is no different from using other digital image manipulation tools like Photoshop, and that human creativity was necessary to craft the right prompts and curate the final award-winning piece.

Controversy over new art-making technologies is nothing new.

The New York Times article shared that “controversy over new art-making technologies is nothing new. Many painters recoiled at the invention of the camera, which they saw as a debasement of human artistry. (Charles Baudelaire, the 19th-century French poet and art critic, called photography “art’s most mortal enemy.”).”

Is text-to-image AI different? Maybe, maybe not. Regardless, human artists are, however, understandably anxious about their futures. Will anyone pay for art or design if they can just generate it themselves? Or are these just new tools that will augment concepting and prototyping, freeing artists, designers, marketers, and more to focus on the more directional components of creation?

Just this past June, Cosmopolitan commissioned art director and digital artist Karen X Cheng to produce the magazine’s first-ever AI-generated cover art — a strong woman shown as an astronaut, based on creative direction from Cheng and the Cosmo design team. The headline and lede on the cover read, “Meet the world’s first artificially intelligent magazine cover. And it only took 20 seconds to make.”  While it took Dall-E twenty seconds to render the image, that bombastic but attention-grabbing claim does not take into account the time it took to refine the art direction or compose the right prompt to achieve the final image.

Cheng documented the process and posted the video on Instagram, which showed the hundreds of iterations of text prompts she typed before coming up with: “wide-angle shot from below of a female astronaut with an athletic feminine body walking with swagger toward camera on Mars in an infinite universe, synthwave digital art.”

Via Cosmopolitan

Cosmopolitan wasn’t the only magazine with an AI-produced cover this past June. The normally buttoned-up Economist deployed a Midjourney-created piece emblazoned with the headline: “AI’s New Frontier.” These examples are significant because they show how quickly digital technologies can go from bleeding edge to market, giving rise to a panoply of complex emotions.

What is the future of text-to-image AI, and what does it mean for artists and designers?

Advances in AI have often sparked concern about the displacement of human workers. While those concerns are legitimate, IBM CEO Ginny Rometty recently said, “If I considered the initials AI, I would have preferred augmented intelligence.”

One way to look at this new technology is that it can help push creative visions forward. Someone putting together a presentation might find they can communicate ideas visually that surpass their artistic abilities. The production team for a video shoot can quickly test out backdrops and props ahead of time. An advertising agency can tweak drafts of a new campaign before having artists work on the final concept.

The reality is that AI design tools are already a part of the creative industry. The Adobe Creative Suite is full of AI-enhanced features. Premiere Pro has proprietary Adobe Sensei AI embedded, allowing automated captions to be created. In Illustrator, the ability to trace and vectorize sketches is powered by AI, as well as the skin-smoothing and other retouching tools in Photoshop’s neural filters.

This compendium of AI use cases from the last several years shows how ubiquitous automated design has become, but it also demonstrates that without creative intervention from human artists and designers, the results of AI-generated design can feel derivative, regimented, and homogenized — making the case that without a human to steer the way, this technology has no intrinsic soul. Perhaps the future is for humans to be captains of creation, with a growing array of digital tools at their disposal.

About the author. 

An award-winning creator and digital health, wellness, and lifestyle content strategist—Karina writes, produces, and edits compelling content across multiple platforms—including articles, video, interactive tools, and documentary film. Her work has been featured on MSN Lifestyle, Apartment Therapy, Goop, Psycom, Yahoo News, Pregnancy & Newborn, Eat This Not That, thirdAGE, and Remedy Health Media digital properties and has spanned insight pieces on psychedelic toad medicine to forecasting the future of work to why sustainability needs to become more sustainable.