Can AI software write better content than humans?

Did you know that in 2016, The Washington Post created Heliograf, an in-house automated storytelling technology, to help cover the 2016 Rio Olympics? Heliograf generated short reports to update readers on the outcomes of various games.

The Associated Press, one of the world’s largest news conglomerates, uses natural language generation software, called Wordsmith by Automated Insights, to publish hundreds of articles. 

Artificial intelligence has taken over almost every aspect of communication, from Amazon recommending products based on a user’s browsing history to Siri answering questions. And AI is evolving at a faster rate than ever: Research shows that by 2025, AI will attain a global market value of $190.61 billion. 

This begs the question: Can AI software write better content than humans? 

Based on data from The World Economic Forum, “By 2025, 85 million jobs may be displaced by a shift in the division of labour between humans and machines, while 97 million new roles may emerge that are more adapted to the new division of labour between humans, machines and algorithms.” 

That said, human interaction still plays a significant role as far as copywriting roles are concerned. 

What lies ahead for copywriters

There are several limitations of AI for content creation. One of the biggest issues with AI right now is its overreliance on patterns and the likelihood that certain words or phrases will appear next to each other when a particular topic is mentioned. 

This means that at present, AI can only come up with a basic draft of factual, information-based content. And still it seems to struggle with understanding anything it is producing. Essentially, AI is simply picking up information that is already out there on certain topics and then coming up with a robotic version of the same.

Content creation with AI is currently limited by these four factors: 

1. Inability to string together long, compelling text

– In most cases, when a human being is developing content, they typically lay the groundwork for a piece in the first paragraph, then come back later to build on that in the subsequent paragraphs.

However, AI cannot work like that. Its understanding of different topics is completely isolated from each other; therefore, it cannot connect the dots that a reader would expect to see.

2. Unable to handle emotion

– AI cannot deal with emotion in terms of anecdotes, storytelling, style, inside jokes, jargon or meta-references. These elements are what gets readers to pause in their tracks, take notice of what they are reading and actually want to read the full thing.



Humans are emotional beings who are hardwired to use feelings to convince themselves of logical decisions, and not the other way around.

3. Unable to work with audio or video

– Most people these days consume digital media in the form of audio or video, and will continue to do so. AI cannot handle that. It also cannot create scripts for this type of media. 

4. Inability to be subjective

– Most content found online is subjective. It compares alternatives or provides recommendations, and has its own pros and cons. So, unless AI is plagiarizing content like a robot, it cannot compare alternatives or provide additional context as to why a particular argument might or might not be valid or legal.  

Areas where AI can help with content creation 

Here are three areas where AI can be of assistance when it comes to creating content: 

1. Preparation and research

– Most long-form content online takes anywhere between five to six hours on average to write. And about half of that time is spent on preparation and research. This is where AI can be a huge help.



With its underlying content technologies, AI can help dramatically cut this time from hours to minutes, and provide ideas and suggestions for how an article should look.

2. In specific cases, first drafts

– In some instances, it may be possible to use AI to create short-form, basic fact-based content for a first draft. However, a writer or editor would be needed to review it, edit it, polish it or, if necessary, add to it.


This would again save a significant amount of time and money, especially if one can work with AI to verify or manually approve an outline before the AI strives to write it out.

3. Pattern-matching (for SEO)

– Pattern matching (a technique where you test an expression to determine whether it has certain characteristics) is great for developing content for SEO. A pattern matching-centered approach to AI-based research can help structure content to be SEO-friendly. 

Takeaway

AI is here to stay. Therefore, when using AI and its underlying technologies for creating content, it is crucial to keep human experts involved at various stages of the process, to verify, filter, modify and remove text to make sure it is legitimate.