AI natural language processing has seen some major advances in recent years. It is now fairly common for sites to use AI chat bots to interact with visitors to handle common questions, direct them to what they are looking for in a more friendly way than search, and identify inquiries that need to be passed to a human staff member. There have also been impressive demonstrations of AI natural language processing technologies being used to write human readable articles.
When an AI is trained on an appropriate library of content it is possible for it to structure sentences in a way that makes sense, based on a fairly small amount of input about the topic for the text from the user. It’s not quite good enough yet to create on topic content reliably enough to generate articles most people would be happy to put on their blog or company website, but it is being improved all the time.
However, with current machine learning approaches, there are still things that AI content writing will not be able to do, and when AI becomes common in terms of web content, this is also likely to have an effect on the internet as a whole, and how certain content is treated by search engines.
Here we take a look at what some of the future implications of advances in AI content writing could be:
The main advantage of AI content writing would be that site owners could produce as much content as they wanted, very quickly, and for far less cost than hiring content writers. While we are probably a long way away from AI content that wouldn’t need to go through some editing or quality control by a human eye before being posted, the cost in time and money of doing this would be lower than that of having humans do the writing as well as the editing.
This may sound great to site owners who invest a lot in content, and also to solo bloggers who would love to have more regular posts and more content on their sites. It also may sound worrying to content writers, and even to other types of writers like journalists and novelists who fear their skills will no longer be needed once AI writing reaches a certain level of reliability. However, there is a bit more to the issue than writing becoming just another job that can be replaced by machines.
With the way that AI is currently trained for natural language processing, it takes into account words and context, but does not cross reference any of what it is writing with facts or timeliness. This means that it is not great at writing about a lot of topics. The obvious example is news or topical articles. Even a topical article about something as innocuous as fashion or home decor trends is unlikely to actually end up matching reality, even if it is perfectly readable, and a news related article about something more significant will be likely to end up generating all kinds of false information.
News is not the only type of content where current AI approaches would not be viable, if you want your content to actually inform people and be trustworthy. Tutorial type content, advice, scientific or medical content, legal or financial content, the list goes on. In some of these cases the AI may write correct information based on what it was trained on, but that doesn’t mean factual and accurate content can be guaranteed. Needing a human being to fact check everything the AI came up with could end up being just as time consuming and expensive as having them research and write the article themselves.
There is also the matter that many sites want their content to reflect the personality of the writer or brand. If you want your content to have a certain tone, for instance witty, friendly, or authoritative, this is not something that can be achieved with the way AI writing currently works, so you may well prefer to stick with human writers who know how to match your brand’s tone and can be creative with things like humor.
Search Engine Value
A concern when it comes to AI being used for content writing is that due to how quick and easy it would be, masses more content would be being uploaded every day. Not all site owners would be scrupulous about the issues listed above around fact checking, and would also use AI to produce content it isn’t really appropriate for, leading to a lot of this content being poor quality and full of misinformation.
Search engines, which keep their user bases by matching people to the most suitable content relating to their searches quickly, would undoubtedly seek out ways to identify the non-AI content and penalize low quality AI generated text in search results.
One area where AI content writing may become common very soon though, is in translation. Machine translations between many language pairs from Google Translate or other systems like DeepL are already good enough to allow individual users to understand websites in foreign languages, even if the translation isn’t exactly what a native speaker would have written.
This presents the opportunity for site owners to make their content available in multiple languages, expanding their audiences and being able to rank in search in several languages too. Translating content doesn’t present the same issues with fact checking as generating it from scratch with AI, and this also doesn’t really threaten the translation industry, as it would mostly be used by people and businesses who otherwise wouldn’t have translated their blogs or catalogs at all. Things like fiction that require more nuanced translation, and things like legal documents that require perfect accuracy, would still need to be done by humans, protecting translation as a profession.
Of course, these predictions are all based on current methods of training AI, and what we can expect in the near future. As technology moves on, AI may be able to rival human writers across all styles of writing – this just isn’t on the cards just yet.