Drowning in AI Content? How to Stay Informed, and How to Be Worth Reading

gcwagner.com | October 2026 | G.C. Wagner

The web is now full of AI-written content, but most of it isn't winning. Readers and search engines still reward accuracy, originality and trust. That's how I filter the flood as a reader, and it's how I write for clients.

Anyone can now publish a 1,500-word article in about 90 seconds. Many people do.

The numbers are big. Graphite's analysis of the web found that AI-generated articles briefly overtook human-written ones in late 2024. A follow-up in May 2026 found the share has stayed near 50% for over a year. Separately, security firm Imperva reported that automated traffic made up 51% of all web traffic in 2024. That's bots visiting pages, not bots writing them, but the direction of travel is the same. This is the "dead internet" worry in real numbers.

AI-written content isn't bad in itself. The problem is volume without value. Most of it restates what the top ten results already say, so it adds nothing for the reader.

You'd expect the flood to take over search. It hasn't. In a study of nearly 31,500 keywords, Graphite found that 86% of articles ranking in Google were written by humans. Only 7% of the articles ranking number one were AI-generated. ChatGPT and Perplexity cite mostly human-written articles too (82%).

Google's own position is consistent with this. Its guidance on generative AI contentsays AI is useful for researching a topic and adding structure. But using it to produce many pages without adding value for users may break its spam policy on "scaled content abuse". Google also reminds us that generative models don't retrieve facts, they predict likely words. So it says to manually fact-check and review all AI-generated content before publishing. That line was added in the 1 October 2026 update.

None of this is new. Google's March 2024 update was aimed squarely at unoriginal, low-quality content and introduced the scaled-content-abuse policy.

1. Trusted newsrooms first. For current affairs I go to outlets with real editorial oversight, such as The Guardian and The Economist. Everything else is optional. The Reuters Institute's Digital News Report 2026 backs this up. Only 22% of people trust news on social media, and 20% trust news from AI chatbots. Trust in the most widely used news brands is holding up better than trust in news overall.

2. Social media is for entertainment, not information. It's built for attention, not accuracy. Fake-news concern is at 62% globally, and I share it.

3. Use AI as a research assistant, not an oracle. I use AI tools with web access to surface sources on a topic. Then I read the sources myself. For finance I go to places with a visible review process:

• Investopedia has a Financial Review Board and requires primary sources.
• Wikipedia isn't perfect. A famous Nature study found it close to Britannica on accuracy, because many editors check each other.
• The CFA Institute writes openly about where AI ends and investment judgement begins.

The pattern is simple: I trust brands that have something to lose if they're wrong.

The flip side is that you want to be one of those trusted sources.

Trust is the goal. Google describes E-E-A-T (experience, expertise, authoritativeness, trustworthiness) as a way to understand good content. It says trust is the most important of the four. It also says E-E-A-T isn't a single ranking factor, so there's no "E-E-A-T score" to game. It matters most on topics affecting health, financial stability or safety, which is why my finance clients need to take it seriously. Google's "Who, How and Why" test is a good checklist. Who wrote it? How was it made? Why does it exist?

Add something new. Google has a patent on "information gain". The idea is to score a page on how much new information it contains beyond what the user has already seen. A patent doesn't prove Google uses it in ranking, but it's a good test anyway. Your content should do at least one of these:

• original data
• first-hand experience
• a clear opinion
• a better explanation than the other pages

Google's own questions say the same: does the content offer original analysis, or does it just rewrite other sources?

Don't over-engineer GEO. Google's new guide to generative AI search says AI Overviews and AI Mode are built on core ranking and quality systems. The same report says Google doesn't need llms.txt files, chunked content or special AI-rewrites. For Google, GEO is largely just good SEO. The academic paper that coined the term, GEO: Generative Engine Optimization, reported visibility gains of up to 40% in generative engine responses. But it tested those tactics on research benchmarks, not on Google itself.

When I write for clients, I ask two questions before anything goes live:

1. Is it accurate? I checked it myself, not just the AI.
2. Does it read well? A real person with a real question would get what they need.

If both answers are yes, you're building the thing that survives every algorithm update: a reader who remembers you and comes back. Search engines and chatbots then find it easier to trust you as well.

AI makes a first draft cheap. What it can't do is care whether the answer is right. That's still the job.