While GenAI represents enormous opportunity for marketers—including elevating the stature of the function within the C-suite and throughout the organization—there will certainly be landmines to avoid along the journey. As with any transformational technology, capabilities and solutions will continue to evolve, likely at an increasingly frenetic pace. In addition, the rules of engagement are being written in real-time, making it challenging to determine what’s effective, appropriate, ethical, and even legal. Here are a few of the most obvious challenges marketers will face.

Data Integrity

When AI is crafting much of our written or image-based content— conjuring up first drafts of emails, letters, reports, blog posts, presentations, and videos on through to computer applications—how will anyone know what is real, if it’s real, and what that means?

Companies are putting AI governance systems in place to guard against systemic bias, inaccuracies, and what are being called system hallucinations—where the AI models confidently make up facts or spew random toxic results. This is because the liability is squarely on the shoulders of the marketer running the prompts. But this governance will always be less than perfect, and additional protections will be needed.

For one, it will become important for AI providers to act with greater transparency and allow the AI tools to cite sources in-line. This will make oversight more efficient and allow marketers to work with data with a higher level of confidence.

There are also ethical considerations from data privacy to truth in advertising that require Human oversight, ensuring that AI-driven marketing adheres to best-practice standards.

The Interactive Advertising Bureau (IAB) is trying to guide marketers in the effective integration of AI and human expertise in marketing. Here is a summary of IAB’s recommendations:

• Use high-quality data that is accurate, relevant, and free of bias.
• Continually assess AI outputs for accuracy, relevance, and appropriateness.
• Update AI models regularly to ensure they remain effective.
• Consider the ethical implications of data privacy and transparency.
• Approach marketing as a human-AI collaboration. 

Copyright and IP Issues

Since generative AI is generating output based on already-existing material, naturally the question is asked:

• Who will have ownership rights to AI-generated content, the brand or some Creator or amorphous set of Creators with hard-to-establish provenance?
• When an AI model produces a new product design or concept based on text that an individual enters at the prompt, who lays claim to that design or concept?
• What happens when the AI model plagiarizes a source that it has been trained on?

Also, now that anyone can create fake images or videos for a few bucks in a few minutes—what are known as deepfakes—the concept of trust has become an even bigger and more complicated issue. OpenAI is attempting to control fake images by watermarking images created by their DALL-E 2 tool. Other platforms will surely follow suit or be compelled to.

A thicket of legal questions will arise, likely drawing counsel into the marketing domain with unwelcome regularity.

Brand Fit and Authenticity

Two decades ago, as we entered the age of digital marketing, we began to see a gradual but steady decline in traditional strategic Brand management thinking. Even classically trained brand managers—who were very disciplined and strategic in other forms of brand activation—seemed to lose sight of the brand strategy North Star when it came to digital marketing. The goal of building long-term brand equity took a back seat to chasing more tangible digital metrics, such as conversions, clicks, followers, shares, and likes.

Now as we enter the age of AI, marketers need to be redoubly focused on the North Star of marketing discipline. AI could very easily squash real creativity and genericize brands, given that it is trained on an existing fact base and cannot reliably extend beyond that fact base.

Since AI content lacks the emotional intelligence, we commonly associate with human creativity, in using it marketers risk muddying the brand positioning unless the process is carefully managed. 

So now in this new age, the generating of content using AI can be so easy that, as they say, even an eighty-year-old grandmother can do it. But generating content using AI that is on brand will be far more complex work.

Companies need to layer brand plug-ins on top of company systems to ensure that the brand positioning and personality are properly executed within the brand’s approved framework.

AI will eventually get to the level where it can develop an entire strategy. But is it appropriate for your brand? Are you going to be able to write algorithms that tie into your brand strategy, hierarchy, architecture? Maybe at some point. But right now, I think we’re still in an experimental phase.

Brian Miske, KPMG

Fake Ads Blanketing the Web

Startups like SpeedyBrand are using AI to create SEO-optimized content on behalf of companies. Their business model is simple: it costs a penny to run an ad, a penny and a quarter can be made on it, send out billions of ads, and you make bank. Meanwhile, the web is so spammed up that legitimate companies cannot break through.

There are even larger concerns.

Many of these synthetic ads are running on fake sites that have been built to exploit programmatic advertising systems. The news vetting company, Newsguard, released a report finding some 400 ad placements from 141 major brands on 55 junk news sites—so the ad revenue is being siphoned off from name brands.

One deepfake video scam running on Facebook shows UK consumer finance champion Martin Lewis appearing to shill for an investment that’s also backed by Elon Musk. Of course, the footage is all fake, and Mr. Lewis’s attempts to shut it down have fallen on mostly deaf ears at Facebook.

Compounding the problem, say researchers at the University of Zurich, people may be more willing to believe information, or disinformation, generated by AI than that generated by actual humans. They asked ChatGPT to create both true and false tweets; they also collected true and false tweets from Twitter. Then they asked a panel to judge the accurate from the fake. Panelists were 3 percent less likely to believe the false tweets that humans had written (making the AI a better liar, essentially). Researchers believe this is because ChatGPT’s tweets were better structured and easier to process, an advantage that will only grow with use. Researchers cautioned against alarm, noting that many things factor into the believability equation. Nonetheless, this suggests that AI will take communication further away from truth than toward it.

Filters

Now entering its fourth decade, the public internet has long remained one step ahead of the filters placed on it. That’s not likely to change. With more than one hundred million people signing up for ChatGPT in just its first three months, there’s going to be a lot of creative mischief. All the more reason to build a shield of integrity around the brand so that discerning consumers can recognize the real thing.

Systemic Biases

Since AI draws on massive amounts of data, avoiding unwanted biases becomes a challenge. This is likely to be AI’s biggest early victory, however. Rooting out systemic biases, to the extent that any human- made machine can, won’t be easy or ever perfect. But it is likely to take place acceptably on the near horizon. 

Speeding this process will be marketers insisting on gaining a God’s-eye view on the data generation process, to better understand the algorithms used to compile and structure the data and ensure the technology is being employed ethically.

Maintaining the Human Touch

AI may well be the most formidable, and even fearsome, tool that humans have invented. But in the marketing realm, fair to say that it represents an operational upgrade and nothing like the “silver bullet” that some promoters claim.

puts it this way:

At Accenture, we talk a lot about AI being the co-pilot of creativity; an amplifier of human talent. Simply put, humans need to do what only humans can do. And AI should make it easier, faster, and more efficient for them to do it. On the front end, it can be used to broaden your views and to give you a wider field to play in creatively. On the back end, it can be used to reduce the burden on time-intensive, laborious activities like versioning.

Jill Kramer, Accenture

Echoing these sentiments is Suzanne Kounkel, global and U.S. chief marketing officer at Deloitte. “The key is to let machines do the work that we as humans aren’t as good at doing, so we can focus our attention on the areas where we excel. For example, machines can quickly and efficiently process large data sets from diverse and dispersed sources, while humans excel at emotions, empathy, strategy, and creativity. The machine can do 80 percent, but then you want humans to take the last 20 percent. I frame it as trying to get to the highest level of consistency without squeezing out the empathy and emotions only humans can bring.”

This “human component” remains an essential part of the equation. Even as AI improves and overtakes humans in one capability after another, that AI will—and should always—serve as an adjunct tool. It may become first among tools in some situations, but only a tool with humans needed to give it direction, provide creative insights, and ensure that its output resonates with target audiences.

Humans are emotional beings. Marketers’ ability to evoke these emotions is at the very core of our success as marketers. Each new version of ChatGPT will be able to write better poetry and no doubt tug harder on the heartstrings, but it will never be human—and that will matter.

Reduction of Creativity

The meteoric rise of generative AI marks both a beginning and ending for marketing creativity as we know it. Large language models such as Bard (now Gemini), ChatGPT, DALL·E, Midjourney, or Stable Diffusion … bring tremendous computing power, speed, and scale to the human act of creativity and ideation.

Jay Pattisall, Forrester

This AI-generated content risks corrupting the artistic expression at the heart of content that effectively catches the eye, sparks interest, and ultimately engages. By simplifying the generation of content to the best the robot can do, these tools risk flooding the marketplace with homogenized identical-looking communications that are soon tuned out, worthless to the originators.

Upcoming Regulations

In Washington, there is genuine concern over AI naturally. Louis Rosenberg, chief executive of AI developer Unanimous AI, has explained to policymakers that AI gives businesses the opportunity to exercise targeted, customized, influence at scale, which, if left unregulated, could be the most dangerous technology for human manipulation that we’ve had to confront.

Many lawmakers agree, and so, tighter government regulation of AI will be an agenda item for years to come. Most marketers recognize as much and are taking pains to adopt policies to protect against the potentially negative consequences of AI. This extends to dealing with AI security breaches from employees using externally sourced AI tools without proper guidance or agreed-upon supervision.

Of Possibilities and Pitfalls

Will we see generative AI used irresponsibly in marketing? Of course, we will. Will we also see in the wider scope an application of AI breaking cancer’s deepest codes? Again, we surely will. So, a healthy combination of both optimism and skepticism about AI will be required to navigate this new age.

Only a small handful of enterprise CMOs are taking a wait-and-see approach, knowing the risks of that. AI is set to truly transform how marketers work; indeed, it is already doing so. Taking charge of the situation begins with prioritizing transparency and accountability, ensuring that data is being used ethically and responsibly in alignment with company values.

Taking such a proactive approach will help CMOs mitigate the very real risks while reaping the benefits of AI technology. It will effectively put the CMO in the position of using marketing strategy to drive business strategy, earning a seat at the big table.

©2024 DK New Media, LLC, All rights reserved.

Originally Published on Martech Zone: Nine Challenges Brand Marketers Face with Generative AI