My blog focuses on Financial Literacy/Money, Business/Entrepreneurship and Technology. Artificial Intelligence is slowly expanding and impacting every aspect of society, the business sector included. One aspect you need to be careful is AI-Generated Marketing. The following contributed is entitled, Why You Should Probably Avoid AI-Generated Marketing.
* * *
AI-generated marketing copy is taking over the internet. The stuff is everywhere, and reading content feels more robotic than ever before. Every word, it seems, is in its proper place, but is it hitting home?
That’s the question that many businesses are now asking themselves. When ChatGPT first hit the scene in 2022, everyone in the industry was excited by what the future might hold, especially for SEO. But over the years, things have changed, and now a lot of professionals are more skeptical about what it is that AI can really do. Yes, it produces buttery smooth content, but is that always what audiences want?
The purpose of this post isn’t to tell you whether content was written by AI or not. Instead, it is to ask a deeper question on whether you should be using AI at all in your marketing copy or any of the other content your business produces.
Here’s why you should probably stick with people:
AI Lacks Authenticity
The first reason to stick with people is the fact that AI lacks authenticity. If a machine is writing your content for you, by definition it means that you’re not.
One of the problems with this is that AI could fail to capture your brand’s tone. It may sound like someone else, or just plain generic, which is what you want to avoid at all costs.
If AI does lack authenticity and isn’t able to mirror your brand voice, then customers will notice immediately. They will wonder why you sound different.
More sophisticated readers will also detect that you have used AI, just by the way things are phrases. While people aren’t always able to determine whether specific passages have been written by AI without a detector, they are pretty good at working out if things have been written this way when considering the broader context.
AI Risks Plagiarism
Another issue with using AI is that it risks plagiarism. The problem with AI right now is that it isn’t really able to move much beyond its training data. Therefore, it can’t usually generate brand new content that hasn’t been used anywhere else. Usually, what it does it bring together ideas that are already out there and express them in its usual, formal, predictable way. This might be okay if you’re brainstorming for CTA button text, but it doesn’t work when producing original content or trying to come up with something new that other sites will link to.
AI Misses Nuance
You may also find that AI misses the nuance of your brand. You may want it to focus on a specific element of what your company does, but that it continues to revert to the industry norm, which isn’t ideal.
When AI misses nuances like this, it can be challenging to train. You have a clear idea of what it should be saying in your mind, but it doesn’t.
When this happens, it can be frustrating, and AI may simply lack the context in its training set to produce acceptable output, forcing you to put up with something sub-par or simply write it yourself.
AI May Lack SEO Control
You may also find that AI lacks SEO abilities when it comes to things like inserting the right keywords or arranging content in the optimal fashion for Google. Systems are great at producing blocks of text, but they struggle with combining multiple instructions, like using Surfer SEO to optimize content.
This issue is problematic because it means that AI isn’t all that helpful for many professionals in the industry. It can write faster than any human on a keyboard, but it can’t do so in a way that is always commercially relevant or sufficiently competitive.
When this happens, it can often increase the workload for marketing professionals. AI produces the content, but then requires extensive editing that can last for hours, taking up a lot of time and often producing results that feel clunkier and less organic than if a human had written it originally.
AI Requires Heavy Editing
Then there are the additional editing issues that AI requires, especially in marketing content that needs to be accurate. Despite three years of publicly-accessible large language models, the issue of hallucinations hasn’t gone away. AIs still say things with confidence that are blatantly untrue.
This problem is also getting worse according to research as more AI-generated content makes its way onto the internet. Newer models are training on the slop produced by older ones, leading to worse outcomes when it comes to truth-seeking. These days, marketers have to check every statistic and truth claim that artificial intelligence makes, simply because the error rate is now so high.
Furthermore, AI can’t always understand complex topics where the original training data is limited. It doesn’t have a good grasp of deep concepts in physics, biology and economics, which is why expert humans still dominate these fields. If it is asked questions about these, it can only supply cursory responses that betray a lack of true understanding at the core of these systems.
AI Produces Low-Value Content
Finally, AI has a tendency to revert to the mean and provide users with low-value content. The copy it produces often tries to please everyone, and winds up pleasing no one.
For example, most AI systems won’t write in a controversial or thuggish way, which is essential for a lot of brands. Furthermore, these systems will also sometimes shy away from certain topics, making it even more challenging to make headway.
The conclusion is that using AI for marketing copy is sometimes helpful, especially in the ideation phase, but it also comes with many drawbacks. AI systems don’t have the common sense or wider nuance that people have. And they struggle to check facts, often spewing nonsense with confidence, revealing that they are more akin to “idiot savants” than real thinking entities.
Of course, all this may change in the future, but it is unlikely to happen under the current paradigm. Large language models may be reaching their inherent limits.
