How the Next Wave of AI Can Transform Research and Insights

Picture yourself in a late-night brainstorm. Usually, you’d call it a day, but now you have an always-on partner: one that can instantly design a campaign, test whether it will work, and suggest how to make it even better. It’s like having the world’s greatest ad designer, behavioral scientist, and strategist on speed dial. That’s the real promise of combining predictive, suggestive, and generative AI.

The Hype vs. Reality

The headlines love generative AI. Slick ad mockups, witty slogans, and catchy images pop out in seconds. And yes, in a way, it feels magical.

The impact of generative AI on advertising is undeniable and growing rapidly. The global AI in marketing market, which includes generative AI, is projected to surge from an estimated USD 20.44 billion in 2024 to USD 82.23 billion by 2030, demonstrating a robust compound annual growth rate (CAGR) of 25.0% from 2025 to 2030. Specifically, the generative AI in the advertising market is expected to jump from $2.72 billion in 2024 to $8.1 billion by 2029, with a CAGR of 24.4%. This expansion is driven by factors such as integrating AI with augmented and virtual reality in ads, increasing use in optimizing cross-channel marketing strategies, and significant investments in AI research and development by leading tech companies.

But looks can deceive. A polished GenAI spot can feel emotionally flat, oddly uncanny, or even trigger backlash when it hits real people. Consider Coca‑Cola’s 2024 AI‑generated Christmas reboot of “Holidays Are Coming.” It achieved massive visibility, yet the audience reaction ranged from mixed to negative. Viewers called it soulless, uncanny, even dystopian. Commentators pointed out visual oddities and a lack of warmth. Coke later stressed that humans were involved, but the public takeaway was simple: the ad felt less human. This is impactful as a lesson in how brand memory is built on emotion and authenticity, not just visual polish. And it told a clear story: ads made with generative AI can fail horribly.

Creative Bloq, a website providing “daily inspiration for creative people,” documented five high‑profile AI ad misfires in 2025 across big names like Volvo, Vogue, Skechers, and Toys “R” Us. The pattern was similar: strong novelty at first glance, then a wave of consumer pushback around authenticity, quality glitches, or brand fit.

This may ring a bell for those who have dabbled in the space of neuromarketing: here, we have long and often found a difference between attention and engagement. You can capture the eye momentarily, but you lose the plot if the brain tags it as inauthentic. Even worse, even a subconscious response that something is “off” can lead to an Uncanny Valley response. This type of response is characterized by an uncomfortable feeling of unease, or revulsion, that people experience when they see something that is almost, but not quite, human-like, such as a robot, CGI character, or doll If anything, the Uncanny Valley response leads people to not only look away but also create negative associations to anything that is presented at the same time, such as a brand.

What the research says lines up with this story. Nielsen IQ’s late‑2024 report found that AI‑generated ads were more often labeled as annoying, boring, or confusing, and they were less memorable than conventional ads. An eMarketer poll in early 2025 reported that roughly two‑thirds of adults felt uncomfortable with AI‑generated ads. Academic work has also flagged trust and manipulation concerns when AI inserts ads into conversational contexts; once people recognize the tactic, trust and perceived integrity drop.

In short, the hype needs a governor, and that governor is predictive rigor and human judgment. Fortunately, a whole suite of AI solutions are now available that do just that, offering a powerful counter-narrative to the GenAI misfires.

Meet the Team 

To truly harness the power of AI in research and insights, we need to move beyond the singular focus on generative AI and embrace a collaborative approach. Instead, imagine a dynamic trio working in concert: Predictive AI as an astute analyst, Suggestive AI as an insightful coach, and Generative AI as a rapid creator.

• The analyst (Predictive AI) forecasts what consumers will do, digging into everything from attention and emotional resonance to cognitive processing and overall brand lift. 
• The coach (Suggestive AI) then takes those data-driven results and applies the best marketing, neuromarketing, and behavioral science principles to sharpen the strategy and refine the execution.
• The creator (Generative AI) produces campaigns at lightning speed, drawing on the precise guidance and rigorous testing results from its AI counterparts to deliver not just ideas, but ideas proven to work effectively in the real world. 

Predictive AI have typically been used to train on eye-tracking data, and then successfully predict where people will look just from the image or video itself. Later models have been able to predict other types of responses, such as engagement, comprehension, trust, and ad/brand memory.

A Suggestive AI in creative design is an AI that can look at your asset, understand scores from test results, and bring together the best in class recommendations for how to improve the asset.

A Generative AI can take the recommendations from the Suggestive AI and implement them into new asset designs. Since this process usually takes only seconds to minutes, it can be done iteratively to boost asset performance during the campaign planning and design process.

Why Does This Matter? 

When the three AIs work together, something remarkable happens. You get a continuous loop: generate, predict, suggest, and regenerate. Each cycle produces sharper results, each iteration stronger than the last. Here’s how it can look:

1. Everything starts with an initial asset, be it human-made or made by generative AI.
2. Predictive AI is the “measurement stick” that cuts through the noise and opinions, and reveals what will truly resonate to that specific asset. Will the right things be looked at? WIll it generate emotional appeal? Will the audience understand the message? Will they remember the ad or the brand, or nothing?
3. Suggestive AI interprets those findings and shows how to improve them. What have we learned in neuromarketing and marketing over the past decades on ad design? What are some dos and don’ts that should be tried out? 
4. Now these recommendations are sent to a generative AI that makes multiple versions of the asset. And this is then sent back to stage 1, and the process repeats. 

In our lab, we have recently observed that 75% of creative variants improve at least 11% in performance after just one iteration. That means if you start with several generated options, pick one of the stronger ones, and refine it again, you can move the needle meaningfully in a matter of minutes, not hours or days.

Adjacent research underscores the power of such adaptive testing. For example, JD.com’s human-based Comparison Lift system (using bandit experimentation) reported 46% uplift in click-through rates across its test campaigns.

Real‑World Use Cases, with Numbers That Matter

To illustrate the tangible impact of this integrated AI approach, let's look at how it is being applied in various industries, where they are delivering measurable improvements and demonstrating the power of combining these AI capabilities.

Generative AI with Reinforcement Learning at Facebook

In one of the largest field deployments so far, Facebook researchers tested a reinforcement-trained generative model called AdLlama on ad text. Over 10 weeks, across ~35,000 advertisers, 640,000 ad variants, and billions of impressions, the system achieved a 6.7% higher click-through rate than a strong baseline. The study shows that when generative output is combined with predictive feedback signals, the result is not just novelty but measurable lift in a hard business KPI.

AI Imagery in Disguise

A large-scale SSRN study analyzed 16 billion impressions and 116 million clicks across over 2 million ad-days. The findings were striking: AI-generated images outperformed human-made ones in CTR, but only when they didn’t look “obviously AI.” When telltale cues like odd proportions or oversaturation appeared, performance dropped sharply. The takeaway is clear: generative output can work better than human creatives, but only if it passes the authenticity test.

Predictive AI as a Performance Forecaster

Quirks Media reported results from applying predictive AI models to hundreds of ads. Cognitive-processing metrics such as “cognitive demand” and “focus” were tested against in-market outcomes. The result: these predictive scores significantly influenced breakthrough ability and brand recall. In other words, predictive AI isn’t just an academic exercise; it can forecast which ads will land, before they hit the market.

The Cautionary Side: Memory Matters

As mentioned, NielsenIQ’s 2024 research offers a warning. When consumers were exposed to AI-generated ads, they often described them as “annoying” or “boring.” More importantly, neuro-based measures showed these ads triggered weaker memory activation than conventional ads. This means that they might grab a click but fail to lodge in long-term brand memory. It’s a reminder that predictive and suggestive layers are essential: without them, generative AI may optimize for surface attention at the expense of deeper persuasion.

The AI Playbook for Campaign Planning

This playbook outlines a dynamic, iterative approach to campaign planning, leveraging the combined power of AI tools to transform your creative process and drive superior results. By embracing this methodology, you can elevate your team's efficiency, precision, and impact, ensuring every campaign resonates deeply with your audience.

1. Start with a precise brief and constraints: Begin every campaign with a clear, detailed brief that defines your objectives, target audience, and any limitations. This foundational step ensures that all subsequent AI-driven efforts are aligned with your strategic goals.
2. Run Generate → Predict → Suggest → Re‑generate in tight cycles: Treat your AI models like capable junior teammates, engaging in rapid, iterative cycles of content generation, performance prediction, insightful suggestions, and refinement. This continuous feedback loop optimizes your creative output. 
3. Assign one job at a time, evaluate outputs, give crisp feedback, and escalate the difficulty: Guide your AI assistants by giving them specific tasks. Carefully review their outputs, provide clear and concise feedback for improvements, and gradually increase the complexity of the challenges to foster their learning and enhance their capabilities. 
4. Keep a human in the loop for cultural fit, ethics, and brand voice: While AI offers incredible power, human oversight remains paramount. Ensure that a human expert is always involved to guarantee cultural relevance, ethical considerations, and the authentic preservation of your brand's unique voice. 

From this quick round, it is clear that generative AI is in dire need of relevant feedback from predictions and insights. It’s kind of strange, but this suggests that generative AI needs more relevant data as input.

What Should You Measure?

This should be solved mainly by the predictive AI solution, as its KPIs should be based on what you value the most. Below are a few ideas for what you should expect to measure success on.

• Leading indicators: attention, emotional relevance, comprehension, brand‑memory potential. 
• Bridge metrics: predicted lift or impact scores that combine those components. 
• Lagging indicators: in‑market effects such as CTR, conversion, cost per conversion, brand lift, add‑to‑cart, and sales conversion. 

The operational win is that you come to market with creative that already carries a higher prior probability of success. Speed without signal is theater. Speed with signal is strategy.

The Opportunities and the Risks

The opportunities are thrilling. With this trio of AIs, you can test and refine campaigns before they ever hit the market. Smaller insights teams suddenly have access to expertise that once required Fortune 500 budgets. Speed and precision become everyday expectations rather than occasional luxuries.

But it’s not without risks. If we blindly trust AI outputs, we risk launching shiny failures. If everyone leans too heavily on the same systems, campaigns could begin to look and feel eerily similar. And if human oversight slips, the nuance of culture, empathy, and ethics will be lost in translation.

Having run research labs and companies for a couple of decades, I can confidently say that AI works best when treated as an assistant, not an oracle. Like a bright but inexperienced employee, it needs guidance, review, and sometimes a push to improve.

The Bigger Picture 

You have probably hear this statement almost too many times, but it still holds true: the shift to adopting AI is less likely about replacing people, and more about augmenting and empowering them. With these tools, insights professionals can spend less time chasing data and more time asking the questions that only humans can pose. The questions of meaning, culture, and strategy. This future of insights isn’t just about dazzling visuals or clever slogans churned out by GenAI. It’s also about harnessing predictive, suggestive, and generative AI as a team.

I’ll leave you with a metaphor: think of this AI trio solution as your own 24/7 council of experts, ready to help you imagine, test, and refine. That’s not a dream for tomorrow. It’s here now, and those who embrace it will shape the next era of research.