Once upon a time, machine-generated content was limited to bad-quality, spammy websites. Recently, however, the technology has improved to the point that content written by AI not only makes sense, but it can also be good quality. At the same time, according to some sources, about 50 to 60% of the traffic on the Internet is made up of bots. Are we headed to a closed loop where a proportion of the content on the Internet is created by and consumed by artificial intelligence?
How effective is AI at content creation?
In terms of machine-generated text, the technology has improved greatly over the last few years. Early attempts – such as those used for article spinning were obviously fake. There was no way it would come close to passing a written version of the Turing test since most of the time the text produced was near enough complete nonsense. And once the search engine algorithms had advanced sufficiently, much of this type of spammy content vanished because it was penalised and removed from the results.
As an example of poor-quality spun content, here’s a very bad version of the first paragraph of my ‘Five ways to engage in lifelong learning post’ followed by the original:
I’m a literacy/tone development junkie so I’m veritably agitated that it’s literacy at work week 2022! This time the theme is ‘learning uncovered’ which has three beaches; learning implicit
Retired powers
new confines
I suppose the first beachfront is the most important because everything additional begins from understanding – and stretching – your eventuality. In this post I’m going to look at five easy and free ways to profit from lifelong literacy.
And the original:
I am a learning/self development junkie so I’m very excited that it’s learning at work week 2022! This year the theme is ‘learning uncovered’ which has three strands:
- learning potential
- hidden powers
- new dimensions
I think the first strand is the most powerful because everything else begins from understanding – and stretching – your potential. In this post I’m going to look at five easy and free ways to benefit from lifelong learning.
(If you want to have a go at this yourself I used Article RewriterArticle Rewriter | Free Article Spinner for Text Rewriting (rewriteguru.com)
But things have developed quickly to the point where machine-generated text is practically indistinguishable from that produced by people. And talking about ‘quick’, I just signed up for a trial of Jasper.ai, gave it the topic of ‘Lifelong Learning’, provided some keywords and received two sample paragraphs in under five minutes from start to finish (which included finding my credit card). Here’s one of them:
‘Lifelong learning is an essential part of self-development. It helps us to increase our potential and to become better versions of ourselves. There are many benefits to lifelong learning, including the ability to stay current in our field, to broaden our horizons, and to improve our problem-solving skills. Lifelong learning also helps us to adapt to change and to be more resilient in the face of challenges. It is therefore clear that lifelong learning is a valuable endeavour that can have a positive impact on our lives. There are many ways to engage in lifelong learning, including taking classes, attending workshops, and reading books. The important thing is to keep learning, regardless of how old we are. Lifelong learning is a journey that can lead us to a more rewarding and fulfilling life.’
It’s not necessarily how I would have written it and possibly the style is better because it misses out on my stylistic ‘ticks’.
And AI is not just limited to creating text. How about Synthesia.io which uses deep fake technology to generate quick and easy training and explainer videos?
Gone are the days of trying to persuade camera-shy colleagues to do a voice-over for a video…
The text generation programs like Jasper.ai or Copy.ai were developed using machine learning, together with artificial intelligence. Essentially, they were given the Internet to ‘read’, and developed their writing from the rules and patterns they ‘learned’ from it.
This leads us nicely to the next part of my closed loop:
How much web traffic is bots?
According to Statista.com over 40% of web traffic is made up of bots – some sources put this figure even higher at between 50 and 60%. These bots include search engine crawlers, machine learning algorithms (like Jasper) as well as malicious bots attempting to scrape data or to expose weaknesses in website security.
Is it really a closed loop?
On one hand, we have content produced by AI, and on the other, an increasing proportion of web traffic consists of bots. Are we going to reach a point where parts of the Internet are a closed loop? AI-generated content being consumed by machine learning algorithms and bots?
Honestly, I doubt that will ever be the case, but it caught my imagination in the same way that Douglas Adams’ Electric Monk did. (This is a character from Dirk Gently’s Holistic Detective Agency. It was developed as a labour-saving device: you have a dishwasher to wash dishes for you, a video recorder to watch television for you – and an electric monk to believe things for you.)
Personally, I’d be happy to delegate social media completely to AI.
How can businesses today benefit from AI content generation?
It’s still not possible to delegate all content generation to AI. Google’s John Mueller insists that purely machine-generated content will be considered spam.
Services like Jasper.ai and Copy.ai are also not cheap and may not be appropriate for very small businesses, especially if the topics are highly technical. If many of your key search terms are low search volume (because of the kind of sector you work in) then these tools are likely not for you.
But – assuming you have budget – there are still ways that machine-generated copy can help create general, top-of-the-funnel blog articles and social media posts:
Probably the most effective way to use machine learning in content generation is as an interactive brainstorming tool. Feeding in keywords and topic ideas, then using the suggestions returned to develop them into blog outlines.
If you find working with a blank page difficult, then using some machine generated text as a starting point may help.
Although spinning content is a very bad idea, repurposing blog articles is a brilliant way to come up with more content. This means taking an existing article and coming up with a new angle for it – rather than just rewriting you are adding value by making it something new. Using AI can help here by giving you new ideas for titles and structures for the revised article.
What next?
Of course, this is just the current situation. Who knows how far the technology will develop: soon writing a blog article might be as easy as saying ‘Write me a 1000-word article about [topic], using these keywords and in a professional tone of voice.’ It would get the work done quickly, would definitely reduce the challenge to come up with a good article – and maybe also take the fun out of it.