Is AI-generated social media copy actually hurting engagement? We ran our own two-month experiment to find out. Using identical webinar topics, the same audience, and the same posting cadence, BrandGlue compared AI-written posts against human-written posts. The results surprised us: human-written content achieved a modestly higher engagement rate (3.98% vs. 3.43%), but the gap was much smaller than expected. Here's what the data means for marketers and how we think brands should approach AI in their social media strategy going forward.

The AI Debate Everyone's Having

If you've spent any time on LinkedIn over the past year, you've probably seen countless opinions about AI. Every week there's a new headline claiming AI is revolutionizing marketing, replacing marketers, or secretly hurting content performance. One claim we kept seeing over and over was that social media platforms were downgrading posts written with AI, causing them to receive less reach and engagement than content written by humans.

The problem was that there wasn't much independent research to support those claims. At the time we kicked off this experiment, nearly all of the studies we found came from the social media platforms themselves, or from broader research on how AI disclosure affects engagement rather than B2B content specifically. While those insights were helpful, they also came from companies actively investing in and promoting their own AI tools. We wanted to know what would happen in a real-world B2B marketing environment without relying on anyone else's data.

So, we decided to run our own experiment.

Our Experiment: AI vs. Human Copy

Over the course of two months, we tested one simple question: Does AI-written social media copy actually perform differently than copy written by humans?

To make the comparison as fair as possible, we removed as many variables as we could. The posts promoted the same webinar topics, were published to the same audience, followed the same posting cadence of two to three posts per week, and supported the same overall marketing objectives. The only meaningful difference between the two months was who wrote the copy.

During the first month, every post was drafted using AI. During the second month, every post was written entirely by members of our social media team.

Because our publishing volume wasn't identical between months, comparing total impressions or total engagements wouldn't have been an apples-to-apples comparison. AI-generated copy accounted for 26 posts while human-written copy totaled 19 posts. Instead, we measured engagement rate, which normalizes performance regardless of posting volume.

The Results

The results were surprisingly close.

AI-written posts generated an average engagement rate of 3.43%, while human-written posts averaged 3.98%. That's a difference of just 0.55 percentage points, or roughly a 16% higher engagement rate for the human-written content.

Going into this test, we expected to see a much wider gap.

Instead, AI came much closer than we anticipated. Human-written copy still came out ahead, but not by enough to suggest that AI-generated content is being heavily penalized by social media algorithms. If platforms were significantly suppressing AI-written posts, we would have expected the difference to be much larger. That tracks with what other researchers are finding, too: Buffer's analysis of 1.2 million posts found AI-assisted content held its own against human-only posts, and Originality.ai's review of LinkedIn engagement found the AI-vs-human performance gap varies quite a bit by industry.

What Our Team Learned

Perhaps the biggest takeaway wasn't that humans performed better. It was that AI performed as well as it did.

This experiment changed the way we think about AI in our own workflow. Rather than asking whether AI should replace copywriters, we started asking where it creates the most value. That's a question we've dug into before — see our full breakdown of what's actually working (and what's not) when we tested AI tools for B2B social media .

For us, AI is best used as an accelerator. It helps generate ideas, organize thoughts, and create strong first drafts in seconds. That saves valuable time and helps eliminate the blank-page problem. It's also a great tool for stretching the life of content you've already created , like turning a blog post into a carousel or a webinar into short clips.

Where humans continue to make the biggest difference is in refining the final product. Brand voice, storytelling, emotional nuance, audience insights, and strategic messaging are still difficult for AI to consistently replicate. Those are the elements that transform a technically correct post into one that actually resonates.

The result isn't AI or humans. It's AI supported by human expertise.

What This Means for Brands

If you're hesitant to use AI because you've heard social media platforms are automatically downgrading AI-written content, our findings don't support that conclusion.

At the same time, we wouldn't recommend copying AI output directly into your publishing platform and calling it a day. Human review still matters, and our test suggests it can provide a measurable lift in engagement.

Ultimately, AI shouldn't replace marketers. It should free marketers to spend more time on the work that matters most: strategy, creativity, audience understanding, and building authentic relationships. That's especially true as AI search reshapes how people discover and evaluate brands on platforms like LinkedIn and Reddit . The human judgment behind your content matters more, not less.

As AI continues to evolve, so will the platforms and the way people consume content. That's why we'll continue testing instead of relying on assumptions.

For now, our conclusion is simple: the future of social media isn't AI versus humans. It's AI helping humans create even better marketing.