AI-driven content writing is becoming increasingly popular with businesses, as it offers promises of time and cost savings. However, this technology comes with many drawbacks that could make it a less attractive option than an experienced digital copywriter. AI-generated content lacks quality and originality compared to what humans can produce; its algorithm may not keep up with current trends or changes in language; and its legal implications are yet to be fully explored. The question is whether AI-written content will be viewed as the content farms of old that simply deluged the Internet with substandard copy in an attempt to beat out the competition in search rankings.
What AI-based writing is, and why brands turn to it
AI-based writing accepts a prompt provided by a user and can return outlines, paragraphs, or even entire articles with the click of a button. There are several reasons why businesses might feel compelled to utilize AI:
- AI-based writing generates content quickly: Compared to sitting down and writing a well-researched and carefully thought out article, AI writes paragraphs in mere seconds based on the prompt you provide to the software.
- AI-based writing is grammatically correct and structured appropriately: The writing that comes out of an AI product will be free of spelling errors and adhere to the rules of proper English, which might appear to be an improvement to those that have worked with low-end copywriters in the past.
- AI-based writing can write on many different topics: AI writing is trained on a corpus of text that encompasses much of the public Internet as we know it. For that reason, it is capable on writing based on a mere prompt provided by the user, and it will attempt to do the rest on its own.
- AI-based writing gives the veneer of expertise: The vocabulary of AI-based writing is generally acceptable on complex topics, and can carry an authoritative voice sufficient to make it appear that it is a subject-matter expert.
- AI-based writing is cheap: As AI generated writing generally costs only a few cents or less per word, it appears easy to pump out a high level of content with little effort or expense.
Why AI-based writing alone is a poor fit for brands wanting to provide an authentic voice
There’s a common theme among the above points listed in that AI-based writing can only provide the appearance of expertise and authenticity. It is true that AI writing can tick all the boxes on a checklist and provide exactly what was requested in the prompt. To an untrained eye, the writing might be sufficient, but both experienced writers and search indexes such as Google know better. There are several critical flaws and telltale traits that distinguish an AI-generated article from a human written one.
AI-based writing is limited in how it can represent language
There are certain difficult, if not currently unsolvable problems in natural language processing in AI, one of which is pronoun disambiguation. Humans are able to make reference and understand pronouns based on their understanding of a world that will consistently elude AI. This is demonstrated through challenges such as the Winograd schema, which would include sentences such as “The box would not fit in the bag, because it was too big.” Humans have implicit knowledge about how objects would not fit into other objects even from early childhood, and any human reader would understand that the box is too large to fit in the bag. AI writing has no concept of the world constructed by the sentence, as it only understands what it has been trained on — and because AI writing is unable to comprehend certain implicitly understood grammatical structures, it will instead simply avoid using them. This significantly decreases the space of the language accessible to the program, and can lead to repetitive, frustrating to read writing.
AI-based writing is detectable by search engines
If you’ve ever used predictive text on your phone, you’ve already got an understanding of how AI-based writing generates text — it analyzes what has already been entered, and then generates options to follow it based on your own language patterns that it has been observing. Authentic human writing will use puns, metaphors, comparisons, and other techniques that a computer analysis would deem unpredictable. On the other hand, AI-based writing will choose words that are highly probabilistic based on the immediately preceding word. Approaches such as GLTR can visually demonstrate just how predictable these models can be, but it’s also evident to humans as well as the writing could be described as “plodding” or “ponderous” since tight, concise writing often involves what an AI program would consider difficult or unpredictable to create.
Google has released a statement that they value experience, expertise, authoritativeness, and trust in their search ranking process, which involves real people evaluating results and determining whether they fit the bill. Because humans are involved in the ranking process, it’s incredibly important to write copy that is not only intelligible, but pleasing to read for those human raters involved.
AI-based writing can open you up to plagiarism
Established writing protocol requires you to properly source or cite material that you obtained elsewhere when presenting facts or arguments. Although there may exist legitimate subject-matter expertise in the model that an AI is trained on, presenting it as your own if you did not produce it yourself can easily land you in legal hot water. There is no defense where you can just say “well, that’s what came out of the program” as you will still be responsible for having it on your site.
AI-based writing might constitute spam
Search indexes such as Google have long penalized sites for practices such as “keyword stuffing” where certain terms are inorganically and repetitiously inserted into copy. Using AI to generate many articles on the same topic will be perceived the same way by search engines, as it is essentially the same practice, only casting a wider net. It is imperative that all of your content is unique and authentic, which cannot be provided by AI writing alone.
Is there a third option?
Search engines live on their user experiences, as users that have to wade through irrelevant content will grow dissatisfied and set out for greener pastures. Google knows this, and just as it has previously adjusted its ranking criteria in the past, it will do so in the future to weed out inorganic AI-content with questionable usability. You don’t just want your copy to be readable, you want it to be enjoyable, and while AI nails the former well enough, it still struggles with the latter.
It’s certainly possible for AI-written content to be useful and suitable in some situations, and in others be completely lacking — and AI can also be used responsibly by a professional with a practiced eye to reap the benefits of higher throughput while also directing it constructively to avoid its pitfalls. It’s quite possible that such a hybrid approach is the best of both worlds, but it still takes a skilled copywriter to know how to harness AI appropriately. Depending on the use case, it’s possible that simply copying and pasting AI-created content might prove beneficial — or it might get you penalized or deranked, if not worse.
Conclusion
AI-based writing might appear to be a useful tool for businesses to generate content quickly, but it is important to remember that AI alone cannot provide the level of detail and authenticity needed in order to rank highly on search engines or convert potential leads into sales. A practiced eye is still required to understand how to be an authentic voice for your brand, and Agency 310 can provide the expertise needed to help you reach your goals.
