Four Ad Giants, Four AI Playbooks

A few days ago, Google Ads quietly rolled out a new feature.

You can now use Imagen 3 to generate ad images featuring real people in the frame. It automatically tells you which image works better for which audience. Performance Max can even run A/B tests on assets automatically.

Pretty interesting. But what really made me stop and think was something else.

The SynthID watermark.

Every AI-generated image gets an invisible marker. Politicians, celebrities, minors, sensitive content — none of it gets generated, period.

Google isn't alone in this. Open up Meta, TikTok, or LinkedIn's ad dashboards, and AI creative is already everywhere. But here's what's interesting: all four are stuffing AI into their ads, yet they're heading in completely different directions.

Let me walk you through each one.

Google: The Automation All-Rounder

Google's approach is brute-force simple.

You throw your assets in, and it generates, tests, and distributes them across Search, Display, YouTube, Shopping, and App — the entire ecosystem. The number of places where a human can step in keeps shrinking.

Imagen 3 handles the image generation. Performance Max automatically adjusts assets. Audience recommendations tell you which image works better for which group. It can even run A/B tests on its own, find the winner, and scale it up.

Powerful.

But watch out for one thing. Google optimizes for engagement, not for creativity itself. It doesn't care whether your ad looks identical to a competitor's — it only cares which one gets the higher click-through rate. And everything is locked inside Google's own ecosystem. Cross-platform alignment? Forget about it.

Who's it for? Advertisers with deep pockets who want scale and don't mind that every ad isn't a one-of-a-kind creation.

Meta: Polishing Existing Assets to the Max

Meta doesn't run fully automated campaigns for you the way Google does. It's more like a photo retoucher and video editor rolled into one.

Images get auto-cropped, backgrounds get recolored. Video length dynamically stretches and compresses to fit Stories and Reels formats. Static images can be brought to life as mini-animations. Copy gets auto-generated in your brand's voice.

And here's a number worth doing the math on. Advantage+ shopping ads grew 70% year-over-year in Q4 2025, driving an annualized revenue pool of $20 billion.

$20 billion. Let that sink in.

But Meta's problem is the same as Google's: all that intelligence is locked inside its own walled garden. You can't take the insights Meta produces and port them to another platform. And it doesn't do fully automated media buying — someone still has to keep an eye on things.

In other words, Meta's AI helps you squeeze every last drop out of each asset, but it won't make strategy for you.

TikTok: The End-to-End Video Factory

TikTok's AI does something different from the first two.

It makes the video for you.

You type in a text description, and it generates a video ad. Scripts are written automatically. Pacing follows TikTok's signature quick-cut style. It can even create virtual characters, so your brand spokesperson doesn't need to be a real person. In October 2024, they launched Smart+ — a direct rival to Google's Performance Max and Meta's Advantage+, also going the automated ad buying route.

One detail I thought was clever: TikTok partnered with Getty Images, so you can pull licensed, high-quality images straight from the asset library. There's also a tool called Symphony Creative Studio — you drop in a product link or some product info, and it assembles a complete video for you.

From URL to finished video — no human in the middle.

That's what makes it genuinely formidable.

The shortcomings are obvious too. Everything it generates serves TikTok and TikTok alone. You want to take that same video and run it on Reels or YouTube Shorts? You'll have to rework it yourself. For brands that only operate on a single platform, it's a killer tool. For marketing teams juggling multiple platforms, it's just one more item in the toolbox.

LinkedIn: The B2B Precision Weapon — No Creative Execution

LinkedIn is taking an entirely different path.

It basically doesn't do creative execution. No text-to-video, no image animation, none of that. All of its AI firepower goes into one thing: getting the right message, at the right time, in front of the right people.

Audience segmentation adjusts in real time. Polls, event promotions — those common B2B engagement formats — all run smoothly. Sales outreach copy can be AI-generated too.

What surprised me most was a tool called Accelerate. Launched in October 2023, by mid-2024 it had already reached half the global market. What does it do? It analyzes your website and LinkedIn company page, then builds out an entire ad campaign for you.

Building a LinkedIn campaign used to take an average of 15 hours. Now it takes 5 minutes.

15 hours to 5 minutes. Think about that.

It also integrates with Microsoft Designer for basic creative customization. But at the end of the day, LinkedIn's ads are still mostly static. You want flashy videos or cool motion graphics? Not happening here.

Who's it for? B2B advertisers. The folks selling software, selling services, chasing leads.

Four Paths — Which Way Forward?

Looking at all four together, one thing stands out.

None of them are building a universal "AI advertising tool." Each one is doing what its DNA dictates. Google has search and a full-stack ad ecosystem, so it pushes toward automation and scale. Meta has the Facebook, Instagram, and Messenger trio plus a mountain of user behavior data, so it pushes toward engagement and personalization. TikTok has short-video content in its blood, so it pushes toward video production. LinkedIn has the world's largest professional identity graph, so it pushes toward precision targeting.

DNA shapes direction; direction defines the blind spot.

Google and Meta are locked inside their own ecosystems. TikTok only serves one platform. LinkedIn doesn't do creative at all. Not a single one can help you manage ads across multiple platforms in a unified way.

So if someone tells you that AI means you can just sit back and let the campaigns run themselves — don't believe them.

AI can generate images, videos, copy, and targeting. It can take you from producing 3 assets a day to 30. It can compress an A/B test from a week down to a lunch break.

But brand voice, cross-platform consistency, and the coherent way you communicate your core message — these are things that still need a human eye, even today.

AI amplifies your judgment — it doesn't replace it. If your judgment is a 1, AI turns it into a 10. If it's a 0, amplifying it still gives you 0.

How much more these four will pour into AI, I don't know. But I do know one thing: the gap between advertisers who know how to wield AI and those who just get swept along by it is widening.

And it's only going to get bigger.