Eight AI Themes to Guide Marketers in 2024
2023 was quite a year for artificial intelligence. Even AI professionals struggled to keep up with the many rapidfire innovations, make sense of the abundant hype, and see through the noise to find the signal that really mattered. The press and pundits hardly helped. Every new development was “mind-blowing,” every downside “existential.” A marketing decision-maker needs more than that to make informed choices about how to best integrate AI into their processes and programs.
As 2024 promises more of the same, I thought it'd be worth summarizing eight key themes I’m tracking for their potential to shape the marketing AI landscape in the coming year — without resorting to hype, hyperbole, or wild-eyed conjecture. They are:
Models go multimodal
AI everywhere (and in everything)
A looming “thin wrapper” bloodbath
AI agents and large action models
Copyright in the spotlight
Misinformation, disinformation, and deep fakes
Responsible AI
The beginning of the end of experimentation
Now, let’s break them down.
Theme #1: Models Go Multimodal
So far, many of the most popular generative AI systems have relied on unimodal models. A unimodal model in AI is like a specialist who's really good at understanding and working with just one type of information. For example, it could be an AI that only understands text, or one that only recognizes and processes images, or another that only deals with sound. Models like this are great because they tend to be really, really good at the one task upon which they’re trained. ChatGPT excels at writing. Midjourney, at image generation.
The downside is that a great piece of marketing content might require a combination of all those things — sight, sound, and motion, all driven by a compelling storyline and script. This would require the use of multiple single-mode tools and a fair amount of human effort to tie their outputs together into a single, seamless asset.
Enter multimodal models. These are AI systems that can understand and work with different types of information – like text, images, sounds, and videos – all at once. Imagine a super-smart assistant who's not just good at reading and writing, but can also understand pictures, listen to and analyze music, and watch videos to make sense of them. For example, Google’s latest model, Gemini, aims for “anything to anything” multimodal functionality, although early demos fell short of the promise and were criticized for being staged.
Nonetheless, the push toward multimodal models is real and “anything to anything” understanding is likely to become table stakes for Big AI.
For marketers, access to applications that use multimodal models promise greater ease of use, better functionality, and a simpler path to creativity when compared to the patchwork of specialist models and apps available today. In fact, the shift to multimodal models may be the innovation required to unlock levels of productivity that generative AI vendors have been promising (but not quite delivering). Like everything riding the AI hype wave, we’ll need to see proof points and real-world deployments to know for certain.
At the same time, with advances in areas like image and video generation, marketers should expect new multimodal AI systems that deliver higher quality, bordering on, meeting, or even exceeding traditional creative methods in many areas. This translates into a variety of applications, including hyper-realistic synthetic humans serving as models, actors, or spokespeople; video production that mirrors the aesthetics of film; faux photoshoots that can easily substitute for the real thing; and more. Every marketer needs to draw the line between authentic and artificial based on their brand, ethos, audience, and outlook.
Theme #2: AI Everywhere (and in Everything)
In 2023, we saw technology companies — software providers, in particular — rush to incorporate generative AI features and functionality into their core offerings. HubSpot introduced its ChatSpot assistant and incorporated AI-powered writing assistance into its marketing suite. Salesforce announced Einstein GPT to integrate public and private generative AI models with customer CRM data. And Adobe rolled out its own proprietary Firefly model, making AI features available in the company’s widely used creative applications. If you sold software, you had to find your AI angle to remain competitive in a fast-moving marketplace.
If the news from the 2024 Consumer Electronics Show is any indicator, manufacturers may now be under similar pressure to integrate advanced AI features. Automakers BMW, Mercedes-Benz, and Volkswagen all debuted in-car chatbots. Sony, Honda, and Microsoft announced a partnership to integrate the latter’s AI systems into a planned smart car. Beyond carmakers, Bosch previewed rearview mirrors with facial recognition technology, Samsung and LG showed off AI-enabled household robots, and Kohler added Alexa to a new line of toilets.
One implication is clear. For marketers, generative AI is good for more than just promotion. It’s finding its way into the first of the four Ps: product. If product innovation is within your purview as a brand manager or marketing decision-maker, it might be time to start thinking about how you can make your (connected) product smarter with AI.
That said, not everything with an algorithm is AI. Neither is every instance of automation. Brands and businesses need to be careful not to overpromise on the extent to which their smarter product truly integrates artificial intelligence.
In fact, faking it rather than making it is so prevalent among marketers touting AI features that the Federal Trade Commission had to issue a warning and promise penalties. Just like any other claim you might make — from taste to fit to product effectiveness — your claims about AI must be true, accurate, and aligned with customers’ expectations.
As the integration of AI becomes ubiquitous, its mere presence in your product ceases to be an advantage. Decisions about when, where, and how to employ AI — whether in your marketing campaigns or your manufacturing processes – must be driven by a keen understanding of your consumer. In short: What unmet consumer needs are you addressing with AI, and how does AI allow you to meet that need better than earlier alternatives?
Theme #3: A Looming “Thin Wrapper” Bloodbath
In generative AI, a “thin wrapper” application is one that adds little value beyond the core functionality of the underlying model. One unfortunate outcome of the past year’s AI hype cycle is that anyone with access to OpenAI’s APIs (or LLAMA or any other GenAI model) could slap a thin interface on top of a GPT and call it a company. To be clear, many (if not most) AI solutions are built on top of foundation models. The issue is the extent to which the application provides substantial, specialized functionality beyond what the foundation model itself offers.
The directory site There’s An AI For That lists more than 15,000 applications, with dozens more being added daily. An analysis by venture firm Sequoia Capital found that few AI applications have nailed their value proposition. As a marketer, you know this can be the kiss of death.
I believe a vast majority of thin wrapper AI startups, along with a significant number of more robust but essentially undifferentiated competitors, will fall by the wayside. This will cause headaches for end users and organizations that have standardized around any given provider for key tasks, functions, or processes. For better or worse, marketing-oriented solutions — including more than a few of the AI writing assistants — arguably fall into the thin wrapper category.
As a marketer, diligence has never been more important if you want to avoid the pain associated with onboarding new technology partners, retraining employees, recalibrating workflows, and even reworking integrations between AI tools and the rest of your martech stack. The right rigor in screening and selecting AI technology partners will go a long way.
Always aim to understand the extent to which the provider delivers meaningful and differentiated functionality you can’t get anywhere else, incorporates a reasonable amount of proprietary features, and has optimized its offering to meet modern marketing organization’s needs. At the same time, it’s worth investigating (even asking about) the company’s financial situation to determine whether they have the resources to stay in business for the length of your contract.
Theme #4: AI Agents and Large Action Models
One of the biggest stories coming out of this year’s CES was the launch of the Rabbit R1 – a boxy little handheld device that uses AI to perform actions on behalf of its user. The company reportedly sold $10 million worth of product within days of its introduction. I can’t say that I’m sold on the idea of a standalone AI device, but the actual AI is intriguing.
The Rabbit R1 runs not on a large language model, but a large action model. So, what’s the difference? A large language model takes a request (your prompt) and generates a response (like text or an image or even computer code) based on its training data. Basically, information in, information out.
A large action model, on the other hand, takes a request (your prompt) and performs an action without any further human intervention. To put this into context: If you ask ChatGPT to book you a hotel room in Chicago, it might recommend a list of hotels but won’t reserve a room because the underlying LLM lacks that capability. But a large action model will actually go ahead and book the room.
Where today’s more common generative AI tools are best thought of as “assistants” that work alongside you, LAM-powered applications might be thought of as “agents” that do the work for you. Even if the LAMs don’t quite live up to the hype just yet, this is a development worth watching over time (with or without a hardware angle.)
What are the implications for marketers then? From a workflow perspective, marketing departments are likely to deploy both AI assistants and AI agents to streamline routine tasks and improve productivity. That much seems obvious (right tool for the right job), but the larger implication is more profound and likely to play out over a longer timeline.
In 2019, I gave a talk about what marketing might look like in 2025. Among other things, I suggested that every consumer would have access to trusted AI systems (agents, if you will) that'd know their personal preferences and purchase patterns well enough to autonomously conduct retail transactions on the consumer’s behalf. While many brands already market to algorithms, the coming reality that machines will increasingly act as unbiased intermediaries between brands and their customers will be one of the most disruptive changes for marketers.
AI investor Jeremiah Owyang wrote about this theme recently as well. Marketers should expect to hear more about AI agents as the year progresses.
It’s more essential than ever for marketing decision-makers to have their finger on the pulse of fast-moving AI developments like the ones I’m highlighting here. At CognitivePath, our AI Advisory services are designed to help marketers sift through the hype to find the themes that really matter, so they can make better AI decisions with clarity, confidence, and speed. Get in touch to learn more about how you can gain one-on-one access to our analysts, unlock all of our research, and empower your organization to win with AI in 2024.
Theme #5: Copyright in the Spotlight
You’re probably aware that the New York Times Company is suing OpenAI and Microsoft, alleging that the companies used millions of the media company’s copyrighted articles to train their AI models. It’s a landmark case, in which the NY Times seeks not only monetary damages but also the decommissioning of foundational models like GPT-4. Nobody knows how this case will play out, but many agree its outcome will shape the future of generative AI one way or the other.
This is far from the only copyright claim to be levied at Big AI. Creators are fighting back against the image generators’ use of their work, as end users, pundits, and AI critics point out the frequency with which these tools output images that clearly ape artists’ signature styles, reproduce movie and television stills with uncanny accuracy, and depict trademarked characters without attribution.
While copyright is important, it’s also just one part of a broader set of concerns related to ownership, intellectual property rights, fair use, transparency, explainability, deep fakes, and even government regulation.
While some AI providers claim they’ll indemnify users against copyright claims, it’s incumbent upon marketers and their agencies to do their own homework. They need to understand what risks they may be taking when they use AI-generated content, insist on better transparency from their vendors, and know how AI affects their own ability to claim copyright or protect brand IP. Marketers should get used to spending a lot of quality time with their corporate counsel, be prepared to pivot, and have a Plan B (C, D, and even E) ready if needed.
Theme #6: Misinformation, Disinformation, and Deep Fakes
Brands will also need to navigate a web (and a world) that's likely to become overrun with AI-generated content — much of it designed to mislead. Misinformation, disinformation, and deep fakes rank among the true risks presented by artificial intelligence. But for the purposes of this article, we don’t need to look beyond the affects on our own industry to understand why this matters for marketers.
Brand Fakes
Taylor Swift doesn’t endorse Le Crueset. Tom Hanks isn’t a dental plan spokesperson. And internet celebrity MrBeast didn’t promote an iPhone giveaway. But millions of consumers saw online ads featuring unauthorized, AI-generated versions of the stars. Meta and YouTube struggle to police their own platforms, even as they remove thousands of AI scam ads.
It’s hard to say whether these fakes are ‘targeting’ the celebs, the brands, or the consumers themselves — probably all of these, to one degree or another. But it’s worth considering how they might damage brand credibility and reputation, and to what extent they upend celebrity endorsements as a tried-and-true advertising tactic.
None of this should take away from the opportunity brands have to innovate with synthetic spokespeople, AI influencers, and other legitimate creative applications of the same types of technology.
Narrative Attacks
Brands need to be aware of — and prepared for — AI-powered narrative attacks. A narrative attack is any attempt by a malicious actor to harm an organization or institution by spreading a false narrative about it. This type of brand safety challenge predates even the internet. But with generative AI, this type of attack is easier than ever to propagate and has the potential to spread out of control. It doesn’t require celebrity star power. Just imagine someone using AI to generate and distribute millions upon millions of false articles that “sound” authoritative (as GenAI often does).
In some ways, this is a 100x evolution of the old social media crisis — those usually weren’t based on false claims, but they did create headaches for brands, sometimes (frankly) well deserved. Dell did put writer Jeff Jarvis through customer service hell and United did break (and initially refuse to repair) Dave Carroll’s guitar. If you remember how even one consumer complaint could go viral and become a movement (back in the day), this is like that but bigger, faster, and harder to control.
AI Clickfarms
Last, advertisers must grapple with an entire underground industry that trades in fake news and aims to attract ad dollars. At last count (Jan 22, 2024), web watchdog NewsGuard had identified more than 600 AI-generated sites masquerading as legitimate news or information sources, except the content is entirely fake, published without human oversight, designed to fool the average user, and set up to siphon ad dollars away from legitimate websites.
On the one hand, these sites might aim to spread mis- and disinformation. On the other hand, they’re designed to attract readership and earn ad revenues spent with programmatic networks. Brands and their agencies routinely review what’s included in the ad networks they buy, creating block lists and such, but because these sites are proliferating so quickly, it can be hard, if not impossible to do this effectively.
Theme #7: Responsible AI
If 2023 was the year of Generative AI, 2024 will be the year of Responsible AI. Amid all the excitement, experimentation, and more than a few warnings about an AI-driven extinction event (bah!), few marketers have paused and asked the hard questions about the risks AI poses. That’s about to change as more marketers turn their attention to Responsible AI — and that’ll be a good thing for individual brands and the entire industry.
In practice, I’d expect marketing organizations to put a premium on guidelines, governance, and standards. Ethical stances and brand safety will be front and center, as it hits home that the marketing end-user — not the model or app provider — is the last (and maybe even first) line of defense regarding the responsible use of AI technologies.
Marketers will deploy AI in ways that deliver results for them while reflecting the values of the company and (this is critical) respecting the consumer’s rights. In the end, 2024 will be the year marketing AI starts growing up — and responsible AI practices will be at the center of this vital evolution.
Naturally, there are specific things marketing leaders can do to make sure AI is implemented responsibly within their own organizations. Begin by defining and documenting clear end-user guidelines. Think of guidelines as “freedom within a frame” – a set of policies and practices that provide three key benefits:
Mitigate the most common and most egregious risks
Protect your employees, company, customers, and brand
Empower your marketers to experience productivity and performance gains
Advanced AI marketers take an even more rigorous approach to AI technology governance, maintaining strict protocols around which systems are used, when, where, by whom, and for what purpose — and carefully tracking and reacting to any potential business, reputation, or regulatory risks that may emerge.
Responsible AI is about doing what’s right. Defining “right” can be a challenging task. Brands might consider establishing an ethics council and even a customer council to help ensure broad, unbiased perspectives factor into the definition and any resulting decisions. Ultimately, your goal as a marketing leader is to implement AI in a way that engenders trust within your organization and between your company and its customers.
Theme #8: The Beginning of the End for Experimentation
We all know that marketing AI success requires more than “random acts of ChatGPT.” So, while experimentation is valid, valuable, and necessary in the early days of any major transformation, it’s important to make the leap toward a more structured, strategic approach.
As more brands stand up and scale up marketing AI programs, ad hoc experimentation will give way to objective-driven, strategic implementations. This will require marketing leaders to:
Establish and communicate a powerful yet realistic vision for the role of AI in marketing processes and programs
Align efforts to objectives
Identify and prioritize high-potential use cases; and design structured, properly resourced plans to pilot and scale AI programs that deliver results.
Invest in upskilling your team and rethinking your workflows.
In short, 2024 is the year smart marketers get serious about AI and set the stage for a long-term strategic advantage unmatched by brands that stand still, stall, or sit on the sidelines.
Conclusion
I’ll be returning to these themes throughout the year, as I work with advisory clients and produce new research to guide marketing decision-makers toward AI-powered success. But at the end of the day, remember, your path to AI success lies not in technology but in humanity. If the past year (or for that matter the past decade) has taught us anything, it's that anything that can be digital will be digital. Anything that can be automated will be automated. It’s the things that can’t be — the tasks, attitudes, behaviors, and attributes that (at least for now) remain uniquely human — that will differentiate your business and your brand.
Put people at the center of your marketing AI strategy, and make sure that the humans in your marketing organization have the agency, autonomy, authority, and accountability they need to thrive in the AI era.
CognitivePath is the only research and advisory firm that focuses specifically on the intersection between marketing and artificial intelligence. Our services include one-on-one AI advisory, hands-on marketing AI workshops, and original AI research. Learn more about how we guide marketing organizations through the AI era.