Generative AI and the Marketing Content Conundrum

I recently read a LinkedIn post on the topic of Generative AI-developed marketing content and the challenges with search algorithms and publishers’ use of AI detectors. One of the comments on the post struck a chord with me. The commenter mentioned they had conducted a great market research study and used AI to write the summary and results. The content fed into the platform was data from the survey - so it was all original work. They then edited the AI summary further themselves. However, when they started submitting the findings report to various publications, it was rejected because it was AI-generated content. 

From the publisher’s perspective, it’s easy to see why they would be hesitant to leverage AI-generated content. AI still has flaws - such as inaccuracies, unreliable sources, privacy and data protection issues, and questions about content ownership. This is a challenge across all industries as AI adoption increases. Where can we trust it and where can we not? What are we to do in the meantime?

I am increasingly relying on AI to generate content for articles, pithy headlines, crisp email content, or summarize and parse large data sets. So the commenter’s anecdote gives me pause and it should give other marketers pause too. I know I’m not alone in this:

• Gartner recently forecasted that by 2025, 30% of outbound marketing communications from large corporations will be AI-generated. 

• Hubspot recently found that 45% of marketers are using AI to write copy for marketing content like blogs and emails. 44% are using it for content QA.

Hence the conundrum facing marketers who are increasingly using AI tools to do their jobs.

Why the Rising Use of AI by Marketers

I can’t speak for everyone, but in the B2B world, leading a mid-size marketing function with mid-size budgets, as I mentioned, I’ve slowly started to  increasingly use AI. And it’s not because my team or I are lazy or dislike writing. As a PR and Communications leader in past roles, I spent my first 15+ years writing a lot of content. I’ve written everything from “scratch” – from press releases, white papers, and blogs to speeches, video scripts and email content. I like to write.

But we now live in a world of:

• Fewer dedicated writers and editors at trade publications and blogs

• An increasing amount of pay-to-play for article placements 

• Smaller marketing budgets (particularly in B2B) for vendors or freelancers, 

• A significant need for more content to support social, email campaigns, and SEO efforts

It’s difficult to keep up with demand. 

AI - The Marketers Little Helper

When you have a smaller team, yet the volume of content that needs to be created continues to rise, AI has been a huge helpmate. 

The most common uses of AI for me have been:

• Providing a variety of eye-catching headlines as a starting point. Many times I don’t use exactly what they give me, but it generally serves as a great brainstorm to spur some fresh thinking.

• Getting me unstuck, so I can move faster. When I’m unhappy with the direction something is going, I’ll sometimes ask for a better way to write a paragraph or additional thoughts that I might consider adding. 

• Checking grammar and giving me that fresh eye.

• Fixing or shortening paragraphs. If you couldn’t tell, I tend to be a bit wordy in my work, so it’s great to an “editor” who can help me be more concise.

  
  

I’m sure not everyone does this, but generally when AI gives me something, I will tweak it further for my personal tone and clarity.

But this is where AI checkers and search engine algorithms  - designed to weed out AI content  – become a challenge for small marketing teams or those trying to bootstrap marketing for small businesses or start-ups.

AI Detectors and Their Impact on Marketing Content

Educators, publishers, social media moderators, researchers and agencies are all using AI detectors and for good reason. In researching for this blog, I wanted to explore how these AI detectors work. As I understand it (all errors are my own), they assess several areas:

• The level of perplexity- humans write with more variation and complexity.. 

• Evaluating burstiness - sentence structure, length, and complexity - all indicators of human writing. 

• Classifiers:  tone, style, grammar, and more are evaluated to look for patterns and assign a confidence score for the likelihood of AI generation.

• Embedding (candidly I got lost a little bit here):  the gist seems to be it’s looking at word-frequency analysis, syntax, and the semantics.

How accurate are these detectors? Some are more accurate than others, but they aren’t 100% accurate. The biggest issue is false positives for people writing original content unaided by AI tools, and false negatives for content fully generated by AI tools.  

So where does that leave people like me who believe they’re writing high quality content with support from AI tools? (Full disclosure, I’m not using any generative AI tools to write this particular blog.)

Google is trying to crack down on “low-quality, unoriginal” articles in search results. In April 2024, they announced they had finished tweaking their algorithms (don’t get me started on keeping up with their algorithm shifts). While I applaud the efforts to filter out spamming content producers - remember the days of keyword stuffing - it’s unclear if they’ve improved the quality content and what algorithms they’re using to weed out low-quality content. Does that include all AI-generated or supported content?

Generative AI is Here to Stay

What is clear is an increased use of AI, particularly generative AI. A survey by Grammarly found that 80% of workers say generative AI improves the quality of their work and 73% say it helps them avoid miscommunications. Assessing the economic impact of generative AI, a McKinsey report stated: About 75 percent of the value that generative AI use cases could deliver falls across four areas: customer operations, marketing and sales, software engineering, and R&D. Another McKinsey report about the state of AI in 2024, found that marketing and sales has led to the adoption of generative AI primarily for content creation and personalized email development.

The likelihood that I’m going to get the budget for a team of writers or agencies to craft reams is slim, and the chances of publishers hiring writers to craft new articles daily are even slimmer. As a result, AI is likely here to stay, so finding the balance of original, quality content – likely aided by support from generative AI – will be increasingly important for publishers and search engines.