AI-generated clickbait articles are likely to hasten the end of search and web publishing
As the latest wave of AI authors emerges, it is feared that they will be perceived as poorly written, riddled with factual errors, and an inadequate substitute for human-authored articles. Google has long faced accusations of bias in its search results, favoring its own web properties and articles written by its own authors over those from freelancers and news organizations. If the search engine starts ranking AI-authored articles, organic search results could be dominated by bots, making search engine optimization (SEO) even more challenging.
Why should Google continue offering organic results if they are written by AI? Human writers can be difficult to work with, often submitting articles late and failing to follow instructions. Some publishing executives believe that using AI to replace journalists is the solution to these problems. Chatbots like ChatGPT and Google Bard can generate text that sounds convincing, enabling the quick creation of multiple articles.
However, the acceptance of poorly written AI-generated content by readers and search engines is questionable. AI-authored articles raise doubts about why readers should visit websites instead of using search engine chatbots for customized answers.
Moreover, why should search engines prioritize organic search results written by their own bots? Google has been accused of manipulating search results in favor of its own products and content. However, showing preference to articles written by Google's own bots would be even more problematic.
Unfortunately, the AI-generated articles I've come across tend to be of low quality. They are filled with factual errors and feature flat or exaggerated writing. Even after corrections, these articles still contain spelling and factual mistakes. Some people may be optimistic about AI-generated articles due to their high rankings on Google, but being briefly in the top 10 search results does not guarantee quality. In fact, it's quite the opposite. The bots can generate large quantities of articles on various topics quickly, making them attractive for SEO. However, this also increases the likelihood of errors slipping past human editors compared to drafts from humans.
Hiring human editors to improve AI-generated content is challenging since the purpose of using bots is to eliminate the need for them. Additionally, human editors are more expensive and may not catch all errors. A potential solution would be to have AI write the drafts and then hire humans to edit them. However, this would require human editors to extensively research every fact and possess expertise in the specific topic. Considering the goal of AI-generated content is speed and low cost, it is unlikely that human editors with subject matter expertise will review each piece.
Another issue with AI writing is that it can result in conflicting perspectives and inconclusive advice, making the editing process difficult. AI draws information from multiple sources, reducing the likelihood of plagiarism accusations, but it also creates an inconsistent viewpoint. While AI authors may improve in selecting facts and using words correctly, they struggle with creating meaning and authority.
Furthermore, leading AI models are showing signs of decreasing accuracy over time. For instance, an AI-authored article about the 25-year anniversary of the Hubble Space Telescope incorrectly stated, "The Hubble Space Telescope was launched by the Atlantis shuttle."
It is unjustifiable to assume that AI writing will improve without considering the associated risks and limitations. We do not extend the same allowances to other technologies, nor do we assume that AI-generated art is valuable simply because some people use AI to create paintings on social media.
The shortcomings of AI-generated content raise serious questions about the future of search and web publishing. Will AI-generated content ever be deemed good enough to replace human authors? If not, will AI articles ever reach a level of quality that justifies publication?
