If AI Writers Are So Good, Why Are Tech Companies Hiring Humans?

AI can write articles in seconds. It can summarize case law, generate practice area pages, and produce thousands of words on demand.

So then why is Netflix paying $775,000 for a storyteller? And why are AI companies like OpenAI and Anthropic scaling human communications teams?

LLMs have made words cheap. When everyone can generate content, you’re only different if you have perspective, depth, and credibility.

This tells us something important

We’ve all heard that writing is now automated. Anyone can prompt their way to scalable content.

And yet the companies building the tools, the ones that understand their capabilities and limitations better than anyone, are investing in people to shape their messaging.

There is a difference between generating words and communicating something of substance.

AI can assemble a competent-sounding practice area page. It can follow a given structure. It can incorporate relevant terms and jargon. It can even sound professional. But it’s only predicting word sequences, not providing perspective. It cannot, on its own, determine what makes your firm distinct or what differentiates you from your competitors.

• Which cases shaped your mission and values?
• What were the judgment calls that built your strategy?
• What’s the most outrageous experience you’ve had at the negotiating table or in the courtroom?
• Which ideas and insights have you earned through years of practice?

We ask our clients questions like these, so our content team can build content that’s unique to their law firm and the attorneys staffing it. The content says something the competitor’s page doesn’t, or puts the firm’s unique spin on a law or case type.

As the web grows increasingly saturated with competent-sounding but hollow content, doing the same doesn’t stand out. It’s why those tech companies are hiring human writers and communications experts. Having a perspective is what sets you apart.

AI is not the boogeyman for writers…

That’s part of the paradox here, and it can be hard to articulate this clearly. There are legitimate concerns for writers about copyright and how AI companies train their models. But when it comes to producing website-ready legal content on its own, the current generation of large language models falls short.

They are tools. Meant, we think, to be selectively used by humans where they can add value. These are not autonomous agents capable of producing distinct, perspective-driven copy for your law firm. Efficiency is value, sure, but when the “efficiency” comes at the expense of product and returns, it’s no longer efficient.

Humans still need to be behind the wheel. Plenty of strong writers use AI productively. They use it to workshop ideas, identify logic gaps, refine outlines, proofread drafts, and even generate early versions of copy. Some build prompts or systems that help them accomplish some of these things, and make their work faster and better.

We’re not arguing that AI can have no place in content production.

…but its use should be carefully curated.

The problem arises when replacing human writers entirely and relying on AI platforms alone to generate finished content at scale.

Any content your law firm produces, whether it’s a practice area page, blog post, social media copy, paid ad, video script, or graphic, should reflect your firm’s ideas, differentiators, and brand. That doesn’t happen with out-of-the-box automated systems. It doesn’t even happen with custom-built GPTs and prompt engineering.

It requires human judgment at the page level. Sometimes that means writing from scratch. Sometimes it means interviewing attorneys and mining for their insights. Sometimes it means building prompts and workflows tailored to a specific firm, then editing and revising the content to meet the required standard.

In many cases, the best results come from a mix of human writers and AI use.

Pure AI content fails to perform in search

An Ahrefs report from last summer analyzed pages ranking in Google’s top 20 search results and evaluated the extent to which they included AI-generated content. They found that most pages (74.1%) included some level of AI, ranging from 1% to 70% AI use. Only 13.5% were purely human-written. But just 7.8% were AI-dominant (70–99% AI use) and only 4.6% were purely AI-generated.

Source: Ahrefs report

That pattern isn’t limited to traditional organic search. Graphite.io also looked into this issue. In May 2024, they found that only 12% of articles or listicles appearing in Google Search results were AI-generated. When they expanded their study in 2025 to examine AI search platforms, the numbers were similar. They found majority of visible content across both traditional and AI search remains human-written.

To be clear, the Graphite.io study did not evaluate AI-assisted content where humans are heavily involved. Similar to the Ahrefs study, they noted that this hybrid approach may be effective.

Source: Graphite.io

So what to make of all this? AI-assisted content is common among high-performing pages, but pure AI content is not. That’s an important distinction because the data doesn’t necessarily suggest that using AI in content production hurts rankings.

There’s a more nuanced story here than “AI content doesn’t rank.”

It’s not the mere use of AI that keeps pure AI pages from performing well in search. It’s that those pure AI pages tend to be thin, generic, and interchangeable. Low-quality human-only content would also perform poorly in search. Meanwhile, the most effective pages involve meaningful human contributions, which aligns with what we see in practice.

Pure AI content appears far less consistently among the strongest-performing pages because human involvement tends to introduce what algorithms strive to reward:

• Specificity
• Depth
• Experience
• Original insight
• Clear perspective

AI can help structure and accelerate content, but humans make it distinct.

If everybody’s sharing prompts and AI systems, isn’t all the content the same?

When agencies talk about “AI content systems,” they’re usually referring to AI-dominant or fully automated production models. We’ve seen plenty of them. We’ve been pitched prompt libraries, turn-key content automation, and AI systems that promise fast, inexpensive content at scale without sacrificing quality.

The pitch is always similar: it’s cheaper, it’s faster, and it performs just as well. And to be fair, it is cheaper and faster. But quality is where it breaks down.

Even when duplicate content issues are addressed through careful prompting or training, the ideas still converge, the angles are the same page to page, and the structure is predictable.

More often, AI content is simply interchangeable.

Here’s a simple test: Open your content and swap your firm’s name with another firm’s name. Does the page still work? If so, it’s generic and interchangeable. There’s nothing distinctly yours about it. No unique perspective, insight drawn from actual cases, or meaningful differentiation. It says the right things and covers the right topics, but it could belong to anyone.

If the pages currently ranking at the top of the search results are already covering the general information, why would an algorithm prioritize another page that says essentially the same thing? Search engines look for signals of value, such as depth and expertise. If your content doesn’t add something meaningfully unique, helpful, and valuable, why would it rank above what already exists?

So, why are tech companies that sell AI investing in human writers?

Because they know all of the above. They know AI can generate language at scale, but they also know that that alone doesn’t create authority, trust, or differentiation.

If you’re investing in SEO while your agency or internal team is producing large volumes of purely AI content, you could be fighting against yourself. Technical optimization can be strong and your site architecture sound. You can build links and improve page speed. But if the content is thin, interchangeable, or lacking depth, it weakens everything around it.

You might even be courting disaster. Google’s Helpful Content Update (HCU) penalties trigger site-wide signals. If a significant portion of your site is filled with low-value pages, that signal can affect all of your site, not just the low-value pages. In other words, poor content drags down the whole SEO and digital marketing strategy.

The companies building AI tools understand its strengths and its shortcomings, and that’s why they still invest in human writers and communicators who know how to use those tools appropriately.

We Do Web is a full-service SEO agency with roots in developing content for law firms. Is your law firm’s content primed to stand out as search evolves? We’d love to talk.

Let’s Do Content Right

Alex Valencia

Alex Valencia is an influential entrepreneur, marketer, speaker, podcaster, and CEO of We Do Web Content, one of Inc. 5000’s fastest-growing businesses in America. His agency implements game-changing content marketing strategies and produces top-ranking web content for law firms, medical professionals, and small businesses nationwide.