AI-Generated Stories can Influence Google Search Results Immensely
Artificial intelligence (AI) is becoming a big part of our lives and is capable of doing the various task, generating convincing text is one of them. Though there are many concerns regarding it but tricking Google is prominent than any other concern.
Numerous blogs, website and marketing spam content can be generated through AI along with keywords embedded in it. The content produced this way is more economical but would not be much meaningful. Still, the search engine would not be able to differentiate it from original content.
Content marketing agency, Fractl showed how AI text generation can be implemented on a large scale for SEO (Search Engine Optimization) Business.
Fractl partner, Kristin Tynski said that AI is capable of producing a massive amount of content which is hard for humans and search engines to discriminate.
Grover, an open source tool developed by Allen Institute for Artificial Intelligence is used by Fractl to write blogs. Tynski said that though the company does not use AI to serve its client competitors are most likely to use it for their advantage.
Tynski explained in a blog post on SEMrush that how almost 10 to 15 years ago a trend of article spinning started. By using tools, the words of original content were changed to make it look like a new piece of writing. Over time, Google and other search engines worked to differentiate the copied content with a lot of filters.
According to her, AI-generated text will be a lot better than spinning articles and will be capable of generating massive content in all possible niches.
An SEO expert, Mike Blumenthal said that spammers will use these AI tools at a huge level, especially considering how massive content is produced at cheap cost. To generate traffic, many news merchants may create fake articles.
Rowan Zeller, who created the Grover tool said that it is possible to detect AI-generated text, particularly at this stage. He along with his team already generated a system that was capable of identifying Grover-generated text, showing 92% accuracy.
Zeller is of the view that AI-generated text has various linguistic and grammatical errors and thus can easily be identified. Instead of forming new phrases or nouns, it uses the same patterns.
Even if a technology is developed to identify it, it has the potential to cause harm. It is not easy to integrate detectors in internet infrastructure. Also, the detectors can also make numerous mistakes and there is no guarantee of complete accuracy of detecting AI-generated content.
Google did not comment on this issue but clarified that it always has and will continue fighting spam.
Proxies like Siri and Alexa are used for web searches and search engines like Google have to show around 3 – 4 more relevant answers. Although, fewer answers may not be enough but they decrease the chances of spamming.
AI text generation is improving rapidly and is expected to set benchmarks. A program that can create content like humans, it will not be a surprise to see it as an AI assistant.
Tynski said that it is possible that in coming years, technology will be so advanced that it will be able to create the content of the level of humans. Once the human-accuracy achieved, it could give humans a tough time, as to why bother humans when there are systems to do tasks.