Marketers Are All In on AI — but Consumers Are Cooling Off

I recently went through a pile of data, and the more I looked, the more interesting it got.

Here's what's interesting: on the marketer side, things are on fire — 56% have already worked AI into real production work, and 70% rank generative AI as the trend to watch most this year. But on the consumer side? Within a single year, comfort with brands using AI dropped from 57% to 46%, and only one in four people trust brands to use AI responsibly.

One side is hitting the gas. The other is hitting the brakes.

This is the one thing from the past year that marketers most need to reckon with.

First, Let's Get the Size of the Pie Straight: How Big a Business Are We Talking About?

Before we talk about AI in marketing, we have to understand how big a pool it sits in.

The global AI market was around $95 billion in 2020. By 2031 it's projected to hit $1.675 trillion. Yes, you read that right — trillion. Nearly an 18-fold increase in eleven years. There was only one dip along the way, in 2022 — a brief pullback right before the generative AI wave — and then it shot straight up.

What about the slice dedicated to marketing? Much smaller, but growing fast. It started at $12 billion in 2020 and is projected to exceed $107.5 billion by 2028. Roughly a ninefold increase.

Think about it: a $107.5 billion market for marketing AI is just one small slice of the overall AI pie. But precisely because it's a small slice, it keeps getting absorbed by the big platforms — as you'll see below, tools like MailChimp, Klaviyo, and HubSpot reach far more users than any standalone AI product.

Machine learning is the largest sub-sector in the entire AI market, projected to land around $552 billion by 2032. That single sub-sector alone is several times larger than the entire marketing AI market.

The World Economic Forum's 2025 survey asked 1,000 employers: which industries believe AI and big data will become core skills between 2025 and 2030? Information technology and telecommunications tied for first, both at 66%. Even the industry ranked last had nearly four in ten saying so.

In other words, this isn't a question of "whether to learn." It's a question of "when."

Money is chasing it too. Of the new unicorns born in 2025, 53% are AI companies — in 2015, that share was only 6%. CB Insights also found that in 2025, one in every five new unicorns is building an AI Agent.

What Are Marketers Actually Doing with AI?

That's the big picture. Now, down at the front lines — which parts of the workflow have marketers plugged AI into?

The numbers from early 2025: 56% of marketers are already using AI in their formal work. Of those, 17% have rolled it out broadly, and 39% use it in specific areas. Another 26% are still exploring. Only 13% have no plans to use it at all.

That's a majority. It means using AI has moved past the "should we?" stage.

So what are they using AI for? Ascend2's 2025 survey ranked the answers —

First place: content creation and optimization, 37%.

Email marketing optimization, 36%.

Social media management and ad placement, 35%.

Content personalization, 33%.

Notice anything? The top four are bunched between 33% and 37% — the gap is tiny. Marketers aren't betting on any single point with AI. They're spreading it across the entire content-related chain.

The influencer marketing sub-sector is even more extreme. Influencer Marketing Hub's 2025 data shows roughly 60% of marketers are already using AI for influencer-marketing tasks — 37.8% to a limited extent, 22.4% broadly. Only 9.5% have no plans to touch it at all. And the number-one thing they use AI for is natural language processing, at 49% — nearly double machine learning. Deepfake technology ranks third, at 24.3% — a number that's a little unsettling.

E-commerce shows a different pattern: it's spread flat. In Klaviyo's 2024 survey, customer support ranked first at 37%, but data analysis, image generation, research, personalization, workflow automation, copywriting... they're all bunched between 30% and 37%. No single link stands out, which means in e-commerce, AI is permeating everywhere — there's no "only used in one place."

There's a number from early 2024 that feels a bit dated now, but it's interesting. At that time, 52% of marketing and media leadership had not used any AI content tools at all. Among those who had, ChatGPT alone accounted for 29% — ten times ahead of the runner-up. This survey was conducted by WordPress VIP, in early 2024. Looking at it today, that distribution has long since been rewritten.

Where B2B and B2C Diverge

On the B2B side, AI works best at the two ends of the funnel.

Act-On's February 2025 survey: B2B marketers say the most effective AI application is audience targeting, ranked first at 43%. Second is analytics and reporting, 41%. Personalization, email, and content creation rank further back, at 35% to 36%.

B2B folks don't use AI to write things. They use it to find people and read data.

As for trend judgments, Mediaocean's November 2025 survey is blunt: 70% of marketers rank generative AI as the consumer trend to watch most for 2026, in first place. CTV and streaming follow at 63%. And the metaverse? 12% — dead last.

The metaverse has gone from "the next big wave" two years ago to the thing marketers care about least.

Using It a Lot, but Trusting It Little

This is the most contradictory part.

Ascend2's 2025 survey asked marketers: how much do you trust the insights AI gives you? Only 13% say they trust it completely. 33% trust it but require human review. 35% say they trust it "somewhat" but in practice still rely mainly on human judgment.

Add it up, and 68% of people are stuck in the middle ground of "trust but verify." Fewer than one in eight is willing to let AI run on its own.

They're careful with money, too. Influencer Marketing Hub's 2024 data: nearly half of marketers (47.6%) spend less than 10% of their total marketing budget on AI-related projects. Only 19% dare to invest more than 40%. The vast majority of organizations are still testing the waters on a small scale.

So what's holding them back?

In HubSpot's 2025 survey, data privacy ranked first — 41% chose it. Training and time investment followed, 39%. Too many tools that don't talk to each other, 34%. Inability to integrate with existing legacy systems, also 34%.

Econsultancy's 2024 data is even more blunt: the number-one organizational challenge with generative AI is unreliable output (including hallucinations), at 35%. The skills gap, 30%. Security risks, 29%. And 19% worry that they're not moving fast enough.

On one hand, they're afraid AI will talk nonsense. On the other, they're afraid of falling behind. Both fears exist at the same time.

When it comes to the specific scenario of social media, Capterra's 2024 data is even more interesting: the generative AI challenge marketers find most vexing is maintaining authenticity, ranked first at 43%. Second is preserving the value of human creativity, 40%.

Neither of the top two is a technology problem. They're people problems.

A Real but Easily Overlooked Fact: Most AI Is Hidden Inside Platforms

Everything above is survey data — it reflects "want to use" and "are using." But what tools are websites actually deploying?

Here's a set of data that looks completely different from the surveys. A detection scan in July 2025 covered tens of millions of active domains.

OpenAI-related integrations are detectable on 41,764 domains. That's the widest reach of any standalone AI tool. Others? LucidWorks on 8,275 domains, DALL-E on 3,038, Clare.AI on 2,368, Gamma Labs on 620, Synthesia on 259.

Sounds like a lot? Compare it to the traditional marketing platforms.

MailChimp: 313,840 domains. Klaviyo: 191,547. Tawk.to: 145,895. HubSpot: 107,974. Tidio: 39,806. Intercom: 22,512.

MailChimp alone is more than seven times all standalone AI tools combined.

What does this tell us?

Many people see the low deployment numbers for standalone AI tools and conclude that "marketers aren't using AI." That's a misread. AI enters marketing in most cases as a feature embedded inside the platforms marketing teams already use — not as a standalone product being installed. Klaviyo's predictive send time, HubSpot's content assistant, Tidio's Lyro bot — these are all AI, but they're hidden inside subscription SaaS and don't get detected as "standalone AI tools."

The survey says 56% are using AI. Deployment detection says there are barely any standalone AI tools. Both numbers are right. They're describing different layers of the same tech stack.

So the most practical advice for marketers is this: before paying for a new tool, audit the AI features in the platforms you're already paying for. The AI that delivers the fastest results is often a toggle switch — not another subscription account.

The Consumer Side: Willing to Pay, but Uncomfortable

Marketers are moving fast. Are consumers keeping up?

About half are. Attest's 2025 survey, across four English-speaking markets, found that 51% of Canadian consumers are open to AI-assisted shopping research, the U.S. 49%, the U.K. 47%, and Australia 43%.

It's even more pronounced when it comes to spending. Appfigures data shows consumer spending on AI mobile apps was $380 million in 2023 and surged to $1.42 billion in 2024. A 274% jump in one year. Consumers aren't just using free versions — they're starting to pay directly for AI apps.

People who use AI apps have a clear profile. Between 2023 and 2024, 89.1% of AI utility app users were male. Every single sub-category skewed male. Education apps were the most balanced, with female users at 31.4%. By age, 18-to-24-year-olds are the largest user group in every category — 65% in companion apps, 56% in education apps.

The core users of AI apps are young, and predominantly male.

But the bottom line of what consumers will accept from AI often doesn't line up with where marketers place AI.

Attest's 2025 survey asked consumers: what do you see as the benefits of brands using AI? Ranked first: faster customer service, 47%. Helping employees do their jobs, 38%. More creative ads, 35%.

Now flip it — what about the downsides? Ranked first: losing the human touch, 59%. Losing jobs and not being able to reach a real person tied for second, 57%. Privacy and security breaches, 43%. Misleading consumers, 40.5%.

What consumers fear most isn't AI making mistakes — it's AI replacing people.

YouGov's January 2024 survey across 17 markets asked the question more precisely: how uncomfortable are you with the following AI advertising tactics? Virtual brand spokespeople replacing celebrity endorsements — 51% uncomfortable. AI-edited product images — 48% uncomfortable. AI-generated product images — 47% uncomfortable. But AI-written product descriptions and slogans — 43% find it acceptable. AI deciding ad placement — 42% find it acceptable.

The pattern is clear: the closer it gets to "that face," the more consumers resist. The more it stays behind the scenes, the more they accept it.

The Trust Gap Is Widening, Not Closing

Qualtrics tracks a particular metric. Consumer comfort with brands using AI was 57% in Q3 2023, and fell to 46% in Q3 2024. An 11-point drop in a single year. And during that same period, brands were accelerating their AI deployment.

Another Qualtrics survey from 2024 hits even harder: only 26% of consumers say they trust brands to use AI responsibly. Three quarters don't.

Brands are doubling down. Consumers are pulling back.

This also explains why marketers themselves are starting to feel anxious. Influencer Marketing Hub's survey shows that marketers worried AI will affect their job security surged from 35.6% in 2023 to 59.8% in 2024. A 24-point jump in one year. Generative AI moved into the daily workflow, and the anxiety moved in with it.

In HubSpot's 2025 survey, the generative AI problem that most frustrates marketers in their daily work is, ranked first, sometimes producing inaccurate information, 43%. Bias, 34%. Off-topic outputs, 31%. Prompts being hard to write, 30%.

So, What About ROI?

We've talked so much about adoption rates and trust levels — but is the money coming back?

McKinsey estimates that generative AI applied to marketing and sales can create about $463 billion in equivalent value annually, with productivity gains equivalent to 5% to 15% of total marketing spend. HubSpot's State of Marketing data points in the same direction: about 80% of marketers use AI for content creation, 75% for media production, and 61% believe marketing is undergoing the biggest transformation in 20 years.

Which scenarios deliver the most results?

Personalized recommendations. Accounts for 36% of digital-experience use cases. It lets small teams do work that used to require an entire CRM team. Brands like Nike have already produced data showing improved repeat-purchase rates through predictive personalization.

Content production. 37% of marketers use AI here — the single highest adoption rate. Writing first drafts, rewriting, localization, batch image generation, video variants.

Customer service automation. Chatbots rank first among digital-experience use cases at 38%, and first among consumer-perceived benefits at 47% — faster service. This is the clearest win-win scenario for both brands and consumers.

Predictive analytics and targeting. B2B marketers rate audience targeting (43%) and analytics and reporting (41%) as the most effective AI applications. Using models to score leads and predict behavior.

Visual commerce and virtual try-on. AI visual tools power try-before-you-buy. L'Oréal has built AI diagnostics and virtual try-on into the purchase path.

What do these cases have in common?

The brands with the thickest returns are all doing the same thing: embedding AI into governed, reusable workflows, preserving brand voice, and keeping humans in the key loops. Chasing speed and volume alone won't earn you anything.

One More Battlefield That's Easy to Overlook: AI Search

You may not have noticed, but AI assistants are now crawling the web fast enough to go toe-to-toe with traditional search engines.

Cloudflare Radar data: in Q2 2026 (April 1 to June 30), across all monitored crawler requests, Googlebot accounted for 27.49%, ranked first. ClaudeBot, 13.87%, ranked second. Meta-ExternalAgent, 12.70%. OpenAI's GPTBot, 10.23%.

ClaudeBot is now the second-largest crawler after Googlebot, having overtaken Meta-ExternalAgent this quarter. The crawlers from Anthropic, OpenAI, and ByteDance each now have traffic on par with Bing's crawler.

What are these crawlers doing? Cloudflare's categorization shows 44.86% are for training, 42.96% for mixed purposes, and only 9.13% for search. Most AI crawlers are stockpiling for tomorrow's answers — the web pages they read today will become the recommendations when consumers ask AI assistants questions tomorrow.

What does this mean for marketing teams? If your robots.txt or CDN blocks GPTBot, ClaudeBot, or PerplexityBot, you've removed your brand from these assistants' data sources. When consumers ask ChatGPT, Claude, or Perplexity to "recommend a such-and-such," you're not in the answer.

AI crawler access is now a marketing setting, not just an IT setting.

A Final Word

String all the data together, and a few lines are clear.

Adoption is no longer the problem — integration is. A majority of marketers are using AI, and generative AI is their most-watched trend of the year. The question has long since shifted from "should we use it" to "how deeply, and where."

The bottlenecks, though, are in people and organizations, not in the models themselves. Reliability and hallucination rank at the top of the challenges list, and data privacy, skill gaps, tool sprawl, authenticity concerns, and job-security anxiety all rank high. To extract more value from AI, the bottleneck is in budget, training, and integration — not in the model.

And the trust gap on the consumer side is widening. Comfort dropped 11 points in a year, and only a quarter of people trust brands to use AI. The closer it gets to "a human face," the more consumers resist.

So what do you do?

Look for the AI features in the platforms you've already paid for, and flip that switch first. Pick a use case validated by your peers — content, email, personalization, or customer service — and pair it with a metric to measure. Keep humans in the loop at the critical points. Check your robots.txt to make sure you haven't accidentally blocked the AI reference crawlers. Then check whether your brand shows up in AI-generated answers.

Running fast doesn't guarantee a win. Running in the right direction does.