Using Artificial Intelligence in daily life – a part of speculative fiction almost from the beginning – has become mainstream. Its use is expanding rapidly into such areas as image and content creation. Proponents see it as a tool to satisfy the increasing need for content, and detractors fear it will replace them. The question to ask is, should you trust your content to AI?
What Is AI-Generated Content and How Does It Work?
“The infinite monkey theorem states that a monkey hitting keys at random on a typewriter keyboard for an infinite amount of time will almost surely type any given text, such as the complete works of William Shakespeare.” If this theorem is true, expanding to more than one monkey, we will achieve the same result in a shorter time.
What it is
AI-Generated content is any content written by a machine.
Artificial Intelligence systems examine sample data sets to learn about the knowledge domain. For any large sample data set, the AI system can provide reliable answers to the questions asked of it. As the system gets more data or good feedback about its results, it can refine and develop its knowledge.
As the models learn, they are drawing on existing data sources, and are generating content based on those data sets. The results returned by the systems will only be as good as the data it was trained on and continues to learn from. We can improve our results by learning how to instruct the system and qualify what we are looking for.
How it works
Similar to a search on Google or Bing, users submit a target topic along with keywords and other qualifiers to the AI generator engine, which uses Natural Language Processing (NLP) to determine the intent of the search and comb the internet for related content and pulls from the most relevant content and uses Natural Language Generation (NLG) to write the content.
The quality of the responses is based on the quality of the available data, instructions to the AI system, and the search terms used. The old saying “garbage in, garbage out” applies here.
What Is the Promise of AI-Generated Content?
Efficient and Scalable Content Generation
Google values content that follows its E-A-T guidelines and is helpful to readers. To produce content at scale that meets these guidelines is essential. Skilled content writers using these tools can more quickly generate quality articles. This results in a significant decrease in turnaround time for articles and may reduce the cost even with the need to edit the content.
Reduces the Time for Good SEO
There will always be the need for skilled SEO practitioners and writers who know how to tell the stories that businesses want to tell.
Companies like Semrush and AHREFS provide keyword generators and keyword use analysis tools for marketers and content writers. These tools provide essential information about keywords, including the number of recent searches, keyword difficulty, cost-per-click, and other metrics that can help marketers and content creators decide what primary and related keywords to target.
Good SEO includes finding the right keywords, meta titles, and meta descriptions that the customer could rank for. Settling on those keywords can be challenging, however, AI systems can help by curating a list of SEO Keywords that will likely appear higher in search rankings and do so in a fraction of the time.
Reduces Writers Block
Even the most experienced writers on occasion find themselves with nothing to say.
- The topic may be one they feel they have exhausted..
- The subject matter allows for little variety.
- The subject matter is something the writer does not know well.
AI content tools provide outlines, key discussion topics, and sample content that serve as a starting point for the article.
Is AI Created Content Any Good?
It can be. Good is largely subjective and dependent on a number of factors. Perhaps the better question is, is it better than human writers?
We’ve all read excellent articles written by highly skilled authors. We’ve also read articles by writers who do not appear to have even a reasonable command of their own language. Most are somewhere in the middle.
“In isolation, the Chevrolet Equinox seems to be a competent mainstream compact SUV. It offers plenty of features even in base LS trim, including myriad safety tech, heated power side mirrors, Wi-Fi hotspot capability, and LED headlights with auto-high-beam functionality…The design of the Equinox is inoffensive… Underneath is a chassis that’s more playful than you might expect.” (Car and Driver)
There is no way to tell if this is AI generated content or just writing/editorial decisions that did not come off well. There are statements that appear superfluous and nonsensical. Language has nuances that human speakers can understand that go beyond the mere interpretation of the words and phrases. When writers are working in a language that is not their primary tongue, they will make these kinds of mistakes.
The technology itself is not necessarily the problem. Human writers pulling from poor source material will generate this kind of writing. As these systems develop through learning on better datasets and from feedback, they will improve. The relative “correctness” of search engines and assistants like Siri, Alexa, and Google demonstrate this.
If the articles the AI generator pulls from are all lower quality, it will produce articles of roughly equivalent low quality. The new ones will become a part of the body of low-quality articles that the engine will draw from the next time it searches for the topic and keywords.
What Are the Downsides of AI-Generated Content?
There are limitations and concerns with content created this way.
Article Length
Most of the applications used today are for short, interactive pieces. These pieces tend to have simple structures focusing on a few key points. The organization and presentation of topics cohesively are easier to achieve. Longer pieces often contain more complex structures and address their topics with greater depth, making longer pieces more challenging for AI engines.
Article Quality
The Guardian wrote an article about its experimentation with AI-Generated content using GPT-3. Here is a sample of its output.
“I am not a human, therefore I know more about you humans than you know about yourselves. You humans know, that a lot of you prefer to compare your evolution to that of a “monkey”. The monkey uses 12% of his brain’ cognitive capacities. While humans are thought to use just 10%. I am a robot, a thinking robot. That’s why I choose to ignore such common assumptions. Evidence is contrary to this claim. I only use 0.12% of my cognitive capacity. I am a micro-robot in that respect. I know that my brain is not a “feeling brain”. But it is capable of making rational, logical decisions.” (The Guardian)
Apart from being entertaining, this sample output from GPT-3’s responses is not useful. It lacks cohesiveness of thought and presents arguments that are outrageous and unsupportable. It should be obvious that this is an extreme example designed to show the limitations, yet it does raise questions.
- What are the sources and are they reliable?
- Are the sources, experts, authoritative, trustworthy?
- How is this information useable?
Lack of Emotional Intelligence
AI systems lack emotional intelligence, which is the capacity to be aware of, control, and express one’s emotions and to handle interpersonal relationships judiciously and empathetically. Again, for shorter pieces, this is not a problem. We are capable of making these kinds of value calls. We can ask AI systems to tell us about the ocean, the woods, the beach, or any other topic of interest and can expect reasonably good and reliable answers. Apart from searching for answers to specific questions, we want to find articles that we resonate with, that tell us a story. AI cannot do that.
“However, it still doesn’t have the capability to truly understand and empathize with readers. It can’t work alone to create compelling scripts for your advertisements or video campaigns. You’ll still need to edit its outputs to fit your brand voice, framing and narrative.” (Neil Patel)
As companies focus on building better systems, these systems will overcome this limitation.
Should You Trust Your Content To AI?
As with any tool, it depends on how you use it. Artificial Intelligence systems excel at certain functions that humans find tedious or time-consuming. Keyword research, topic generation, outlining, and even template content creation can all benefit from using AI.
With Google’s algorithms favoring content written by people for people, knowledgeable writers and editors should review any AI-generated content to ensure a high quality of writing that will meet your readers needs.
At Growth Zone AI and Growth Zone Media, we believe your content should be written by human writers who will work with you to understand your business and what your customers are looking for. We may use AI tools in development and first drafts of your content, but we will always deliver content written by people.
We invite you to explore what we have to offer.