What’s Next for AI and Marketing?
Since its launch in November 2022, ChatGPT, the chatbot developed by OpenAI, has been inescapable. Across the business world and society as a whole, people have been equal parts amused by its capabilities, excited for its potential and impressed by its iterations like GPT-4. They’ve also been fearful, with many speaking up about the technology’s safety, carbon footprint and ethical implications.
The fervor surrounding the new AI tool culminated in an open letter, signed by over 10,000 people at the time of writing, calling for a moratorium on AI development. Though the letter is unlikely to result in a full-scale shutdown of AI projects, it underscores the divide between proponents and skeptics—and at a time when geopolitical and economic uncertainties are injecting more caution into the next phase of AI’s evolution.
Given these realities, we’ve asked industry experts to weigh in on the use cases, possibilities and limitations of generative AI across their sectors of marketing and offer recommendations on how to proceed. The consensus among them, of course, is that along with the advantages come significant risks, and use of the technology without human oversight is irresponsible. It’s clear, though, that the industry is on the precipice of something big, not to be ignored and it’s best, for now, to tread carefully.
Media buying
From expediting research to streamlining ad buys, AI-based solutions can help marketers make the most of their valuable time and resources in day-to-day tasks. As with all AI, finding the right input to ensure a particular outcome is paramount. When conducting research through tools such as ChatGPT, marketers need to set clear parameters: Define how technical the responses should be, a clear subject matter and what the research will be used for. And don’t forget to fact-check.
More sophisticated AI solutions for ad campaigns, meanwhile, can assess which ad formats and mediums are best for achieving business objectives and optimize media buying accordingly. Marketers can specify that they want to purchase inventory that boosts engagement or conversions, for instance, and AI can use previous successes to inform future buys. —Sam Matharu, director of analytics, marketing science EMEA, Xaxis
Processes and workflows
We’re all being challenged to do more with less. But recession or not, success in media is all about the execution of the task at hand. Implementing automation and AI into existing processes and workflows should be seen as an opportunity for growth and expansion of skills.
One area we’re focused on for integrating AI is AdOps, which traditionally is a highly manual and repetitive process and a perfect candidate for automation. Automation enables a reduced order-to-cash process time, allowing for quick invoicing and smoother cashflow; faster revenue reconciliation means a more cash-positive business and a faster ability to reinvest in new business initiatives or new hires. —Jay Kulkarni, CEO and founder, Theorem
Branding and packaging
Using OpenAI’s GPT-3 and DALL-E 2, we let AI create the packaging and product descriptions for seven trendy new products from scratch. From wagyu to oat milk, each product was tested with a nationally representative sample of 300 U.S. consumers to gauge the believability and effectiveness of concept and branding. Each product was graded on a 1-10 scale in seven key measurement areas: overall appeal, likelihood to purchase and others. Results are measured against Zappi’s established norms for product performance, which we’ve gathered by testing over 100,000 products to date.
In short, the AI-created products greatly underperformed with consumers, ranking in the bottom 33% of every product ever tested on our system. Despite aesthetic designs and descriptive language, consumers found the products disjointed and, in some cases, potentially dangerous or irresponsible. It wasn’t all bad, however: The AI-designed products performed well for uniqueness, signaling AI’s potential as an idea generator. —Steve Phillips, CEO and co-founder, Zappi
Thought leadership
As more people start leaning on ChatGPT, we expect to see a lot of subpar thought leadership getting pushed out. This presents a real opportunity for those with genuine points of view to stand out and build credibility for themselves and their companies, and to use AI as a supporting tool to do it faster than ever before.
What makes great thought leaders compelling is the unique perspective they’ve built from years of distinct experiences and a willingness to share it at moments when their business, industry trends and customer needs intersect. An overreliance on AI for thought leadership content creation and contributed commentary robs it of the thing it needs to stand out—originality. AI should absolutely be incorporated as a tool for creating efficiencies, but there are still real limitations in what it can deliver and risks in trusting what it generates, especially because OpenAI hasn’t provided much transparency into the data set the chatbot was trained on. —Brendan Shea, svp, content, INK Communications Co.
Cultural bias
With 1 out of 4 Gen Zers in the U.S. being Latinx, a lack of representation and expertise across general marketing teams and the democratized power of AI at our fingertips, brands and marketers will be inclined to leverage this technology to find efficiencies to engage with Latinx audiences. I cringe at the idea of any marketer crafting multicultural marketing strategy around AI-generated content and justifying it as a cultural source of truth.
When it comes to the multiculturalism of Latines, AI is unable to reflect the diversity of our ambicultural, language-fluid experiences—but more dangerously, its outputs can become an enabler of stereotypes and cultural and linguistic insensitivities. Marketers need to be aware of the technology’s elevated bias when it comes to the Latinx audience, and they need to think of culture-filtering workflows and strategies as they navigate AI adoptions. —David Velez Mejia, executive strategy director, Remezcla
Video and audio
On the marketing side, AI-assisted video creation is helping brands customize content to specific viewers and capture attention in cluttered media environments where visuals and sound trump text. For example, sports publishers are producing videos that integrate player and team stats updated in real time. Similarly, finance publishers like Bloomberg are using AI to showcase the latest market moves. When AI-driven video is implemented, the scope of what’s possible in the creative process expands to what creators can conceptualize and describe rather than what is most practical or cost-effective.
While many took notice of generative AI’s text-to-image capabilities, text-to-3D and text-to-audio have pushed the technology into a new phase. Text-to-video is a major development as well, with companies like Meta and Google developing software in the space. Some capabilities of the tech are creating live portraits from photos converted into realistic talking head models and creating stylistically different versions of original footage for publishers to iterate on successful content. —Dor Leitman, svp, product and R&D, Connatix
Social media
AI is fundamentally changing the way we interact with the internet and on social media. In the coming years, we’re going to see social following become obsolete. With AI powering what content gets seen, the number of users following a brand or creator will have a fraction of the weight on performance it has had in the past. Competition for eyeballs will skyrocket, even for legacy accounts. Speed and risk will become huge factors in achieving success, and because of this, content production quality will decrease. Cell phone footage and less-produced videos will be the new norm due to accessibility—great content is agnostic to the quality in which it’s captured. —Geoff Gates, former head of social media, LA Lakers; current creative director, social strategy and content, Boathouse
Influencer marketing
AI is already helping brands discover creators as well as analyze content to identify good fits for partnerships and how well the content would perform. It can also analyze audiences and recognize fake followers that will not result in quality engagement for brands. Once content is live, AI tools can evaluate campaign metrics and ROI or assist with the campaign itself, automating tasks throughout the process including contract negotiation. There’s also the case of AI-generated influencers themselves, which can potentially give brands more control over campaigns—as long as the partnership lands appropriately with the target audience.
While AI can do great things for influencer campaigns, at what point does the use of AI tools affect the authenticity a creator’s audience relies on? Creators build their followings, big or small, based on trust. Brands looking to partner with creators depend on that same level of sincerity, since it provides credibility to product reviews and recommendations. —Ali Fazal, vice president of marketing, GRIN
Personalization
Over the past few months, there’s been a significant shift in mar tech from marketing automation to marketing autonomy. An autonomous marketing framework has three essential elements—data, decision-making and delivery—and artificial intelligence is the main player in each. Personalized data triggers AI to initiate actions. Decision-making engines handle segmentation and forecast behaviors, ultimately delivering more constant and customized communications to customers.
We’ll see an even deeper integration of OpenAI into creative processes. However, I still see OpenAI more as a productivity tool than a content creation engine. According to Accenture, AI can increase productivity by up to 40%—though tests on our AI email generator proved to be six times faster at creating newsletters, slashing the average time from 19 to 3 minutes. The productivity potential of AI tools may be underestimated, and we are yet to be impressed by it.
What’s left to marketers in the era of autonomous marketing, then? Strategy ideation and creative choices. —Aleksandra Korczynska, CMO, GetResponse
Programmatic
For our AdOps team and possibly others across the ad-tech industry who use it, ChatGPT has been a key resource in generating code for Meta/Facebook pixels using a developer tool link and analyzing competitors’ websites to help create new marketing strategies with a competitive edge. In terms of ChatGPT’s impact within programmatic, that could be more challenging to predict due to the nascency of the tool. Audience and channel planning will still require human insight and optimization. However, there’s potential for a programmatic application to provide optimization recommendations within the demand-side platform during a campaign’s flight. —Liza Bortnikova, COO and co-founder, AI Digital
Search
We don’t use Google search the same way we used to. The average attention span of internet users is 2.5 seconds: When looking for a quick result, many of us scan Instant Answers, the zero-click summary that Google generates, rather than clicking the links provided. Now, ChatGPT is offering an even simpler solution; users can ask the AI chatbot questions and receive detailed, tailored information that directly answers their query.
Microsoft has invested $10 billion in ChatGPT and plans to integrate the AI software into its existing search engine, Bing, whose app saw a tenfold increase in downloads after this announcement. Google has now opened the waitlist for its own AI chatbot Bard. As more users take advantage of the benefits of AI-integrated search, brands stand to gain less traction from SEO and paid Google ads as traditional search engine usage decreases.
Brands must overcome the migration away from Google search due to AI software with a social-first advertising strategy that includes brand campaigns on TikTok. Creating searchable videos that inspire user-generated content will spread awareness of your brand among a Gen Z audience, even as fewer users turn to Google to discover new brands. —Chris Kastenholz, CEO and co-founder, Pulse Advertising
Ecommerce
Retailers can use AI to help answer customer FAQs or accelerate customer support responses. They can also use AI to conduct sentiment analysis on reviews to gain better insights into their products, allowing them to pivot business strategies and approaches to increase sales. While the impact of AI technology is evident, retailers must be highly attentive when checking responses to verify accuracy. This is especially crucial when working with global customer bases and across different languages.
It will be interesting to see the percentage of consumers that turn to AI for product searches and recommendations compared to search engines and ecommerce sites. Retailers should prepare for changes in consumer behavior and the rise of a new channel to add to their marketing mix. There will likely be new industries created around organic and paid placement in the results of these models, much the same way the advertising and search engine optimization industry evolved with the evolution of search engines. —Stephen Curial, CTO, Jungle Scout
Research
I was looking for some textbooks on international advertising and asked ChatGPT if there were any new books that had been published recently. It gave me a list of titles, complete with author and publisher, that got me excited. They were perfect for my class and, surprisingly, I had never heard of any of them. Better yet, they had all been published in the last year. I was thrilled…that is, until I tried to find the books. They literally did not exist. I kept asking ChatGPT how to find the books; finally, it told me: “I apologize for the mistake in my previous answer. It seems that the books I mentioned may not have been published or may not be currently available. As a language model AI, I rely on the information that I was trained on and there may be instances where that information is not up-to-date or may have changed.”
This is not an isolated incident, but rather a deeper problem with AI as it stands today. A recent New York Times article confirmed that AI models “are prone to what AI researchers call ‘hallucination,’ making up facts that have no tether to reality.” That’s not always a big deal for me, but it’s a huge deal if you’re making recommendations to a client or brand based on information that does not exist or has been warped by the software. Long and short, unless you know the data it has been trained on, the data it has access to and how it is organizing that data—and you don’t—you should look at the results with some skepticism and double-check their correctness. —Brian Sheehan, professor of advertising, Syracuse University
Data privacy
The use of AI to create personalized content through third-party platforms raises concerns about privacy. The algorithms need data, either from the brand or the consumer. Are consumers ready for an AI to use their personal images and videos to create branded content? Until we have more transparency on data practices, as well as the storage and sharing practices of these platforms, brands should consider the risks before incorporating them into their content creation strategies. —James Turner, founder and president, Delineate
Copyright infringement
Regulators have yet to make any decisions on whether AI-generated images infringe on copyright rules—but this decision will come soon enough. Cryptocurrency offers a relevant comparison: After the technology had been around for nearly a decade, key players were suddenly being fined tens of millions of dollars by the SEC. While a creative may want to leverage an AI-generated image in a social campaign on a whim, it isn’t worth the potential future risk. Brands must protect themselves from costly, lengthy litigation and the reputational risks that might come with jumping the gun.
The root of the issue comes down to the creative, whose work enters the algorithm and becomes public domain without their consent. In the 1990s and early 2000s, tools like LimeWire and Napster allowed individuals to pirate music with a few clicks and some patience. By the 2010s, Pandora and Spotify delivered an experience people were willing to pay for, so the market returned to equilibrium. While it may take time, the same will occur with AI.
Most brands are willing to pay for work, but technology companies must make it possible to do so. It’s crucial that these companies do their part to provide transparency and compensation around AI-generated images. Regulators must partner with industry experts to develop clear guidelines for their creation and use, as well as rules for ensuring that artists are fairly compensated for their work. —Analisa Goodin, founder and CEO, Catch+Release
Education
We’re at a tipping point in education, requiring us to think differently about what and how students learn. Enter generative AI and GPT. They have a lot more processing power than our students’ brains. They can recognize and recall for us, without much humanity. And that’s where our students can meet the technology. In a marketing course, AI can collect data while students find insights. ChatGPT can write segments while students map them to a customer journey. AI can measure marketing strategy while students interpret the impact of the strategy.
The mar-comms professional of the future needs a relentlessly curious mind, which is an inherently human attribute. We can free students from the cycle of memorization and regurgitation to solve problems and allow them to be what they are best. —Jacqueline Babb, senior lecturer and director of the Integrated Marketing Communications program, Northwestern University