How to Make AI-Generated Content Sound More Human

The evolution of storytelling continues with AI-generated content

Stories have always been part of how humankind has passed on information, and they’ve played an important role in social, political, and religious discourse. As societies evolved, so did the art of telling stories. Today, we’re entering the era where artificial intelligence (AI) is part of our storytelling landscape, with both brands and consumers integrating AI-generated content into their lives and workflows. 

When people write, we draw from our personal experiences and emotional landscape to enrich our storytelling. Human speech is typically filled with quirks like slang, idioms, and cultural nuances, and what we say often depends heavily on the context. AI can’t do that (yet). Which begs the question, what makes AI-generated content resonate with people? What makes us sound human? Should AI-content even aim to sound human?

Can you spot AI-generated content? 

Research suggests that between 30% – 50% of consumers (depending on geographical location) can spot AI-generated copy. But is that a bad thing? Apparently, although half of Americans can tell it’s written by AI, we can’t easily identify any differences in the credibility and trustworthiness compared to human-generated content. And frequently, people even perceive AI-generated content as more engaging, even when they know AI wrote it. 

Research by Markowitz et al. (2024) used a text analysis tool to analyze the difference in hotel reviews written by humans and generative AI. The journal article suggests that AI-generated text sounds more analytical, is more affective, more descriptive, and less readable than a text a person wrote.

Why is that journal article relevant to your enterprise content supply chain? Because search engines primarily reward content that’s designed to give people reliable information. Useful content that’s created for people rather than for search ranking purposes. For content to do the job you intend it to do, it needs to be genuinely useful, sound human, contain accurate information, and be easy to understand.  

The issue with trustworthiness is that just because your readers can’t detect the difference between credible or not doesn’t make it automatically great content. LLMs continue to generate erroneous errors and perpetuate biases. Which is why it’s up to regulatory bodies and enterprises to put guardrails in place to make sure that effective content is also ethical. 

So how do you make sure your AI-content sounds human, is accurate and genuinely helpful? 

Tips to make AI-generated content sound more human

For all the worry about robots taking over the world, they can only do that as much as we hand over our agency to certain technology. So our first tip to making AI-generated content sound human, is to keep humans involved! Use AI as a co-writer, not your primary authoring tool. Collaborative writing has been a technique used by multiple content creators as a way to combine skills, knowledge, and creativity to synthesize original content. When one of these contributors is an AI, it changes the dynamics of creation and ideation fundamentally. We recommend a book called “Co-Intelligence: Living and Working with AI” by Ethan Mollick for a deeper read on approaching AI with a collaborative mindset. 

In the meantime, here are more tips to make AI-generated content sound human. 

1. Write conversational prompts: The best way to get human-sounding text from an LLM is to speak to it like it’s human. Yes, it’s kind of weird at first. But in this interview with Mollick and Forbes, he sums it up as “treat artificial intelligence like a person, but tell it what kind of a person it is.” meaning you have to get good at prompting in a detailed but conversational way where you’re telling it what to do like it was a person, and correcting it as if it were a person. 
2. Make editing inevitable: In the Markowitz et al., (2024) study, researchers claimed that AI-written text has more positivity bias, more emotive language, and is more linguistically complex, making it less readable. Edit AI-generated text to maintain authenticity and build consumer trust, by making sure your claims are accurate, well-cited, and clear and easy to understand. 
3. Base everything on your best content: Ground and tune your LLM off your best content, and in alignment with your enterprise style guide. Need help deciding what content to use to ground your LLM? Use our checklist to guide you to prepare your source content for the highest quality LLM-output. 
4. Keep content on-brand with content governance: AI-generated content comes with the risk of publishing content that creates an inconsistent customer experience because it’s not accurate, or doesn’t apply your company’s tone, voice, style, and terminology consistently. That’s why if you want to make AI-generated content production scalable, content governance guardrails are non-negotiable. 
5. Use quality gates: As part of your content governance guardrails, set up automated quality gates at different stages of your content supply chain that automatically block off-brand content from ever-being published, and send it back to writers for review.

Acrolinx: Keep your AI-generated content sounding like you

The trick to effective AI content is relatively simple: Do the basics well. Just as Google’s systems identify content and rank it in terms of helpfulness, your AI-generated content is similarly scrutinized. As with human-generated content, your LLM has to create content that demonstrates experience, expertise, authoritativeness, and trustworthiness, or what Google calls E-E-A-T. The challenge, of course, lies in understanding what the content needs to resonate with your audience, and how to guide your writers to make these changes.

Does it need to sound human to do those things? Not necessarily. Research already suggests that AI-generated content resonates with humans regardless if they know it’s generated by AI or not. What matters is that it’s representing your brand with a unified voice that’s distinct and recognizable to your audience, and that your writing is accurate, clear, readable, and consistent across channels. It doesn’t exploit consumer’s trust, and adheres to both industry-relevant regulations and emerging government legislation around using AI ethically. 

Acrolinx keeps your AI-generated content sounding like you. Acrolinx is an AI-powered content governance software that captures and digitizes your style guide to make your writing standards, standard. It governs new and existing content written by people and generative AI. Whether your company has written 100,000 words or billions (like our customer Microsoft), Acrolinx makes sure each one reflects your style guide. Customers enjoy massive efficiency gains without sacrificing standards through AI-powered live writing assistance, automated reviews and quality gates, and analytics comparing content quality with performance.

However you tell stories and whoever you want to sound like is your choice. Speak and write like the brand you want to be known for, by your own standards, with Acrolinx.