Monday, 22 January 2024

Google search might be getting worse - and AI threatens to ruin it entirely

A study by German researchers has confirmed what many of us have long suspected: Google search really is getting worse.

A team of researchers led by Leipzig University, Bauhaus-University Weimar, and the Center for Scalable Data Analytics and Artificial Intelligence, have conducted a year-long research experiment to answer the question "Is Google Getting Worse?", and the answer was essentially a resounding ‘yes’. The study wasn’t specifically confined to Google, also analyzing results from the Bing and DuckDuckGo search engines.

The researchers specifically examined product review search terms, meaning that we shouldn’t immediately panic: searches for short questions with factual answers still return accurate results, for example. However, the study did find that across all search engines, “higher-ranked pages” frequently showed “signs of lower text quality” - specifically, SEO spam.

Seek, Engage, Obliterate

For the uninitiated, SEO stands for ‘search engine optimization’: the process of optimizing the content of a web page to secure better rankings in search engine results. This process has evolved significantly over the years; gone are the days when you could simply slap the product name into an article 100 times and lock in the top search result spot. Google works tirelessly to optimize its own search algorithms to combat spam and deliver the best results for users, so the demands of SEO are constantly changing.

SEO has become a dirty word among many online media outlets - trying to keep up with Google’s ever-shifting algorithms often sees sites throwing everything at the wall to see what sticks, and sometimes results in low-quality content being ‘rewarded’ by search engines because it manages to tick the right boxes. SEO has become a cottage industry of sorts, with companies promising (sometimes falsely) to ensure that your site will be the top-ranked result for the relevant search terms. The need for good SEO can come at the expense of content quality itself, a frustrating paradox since high-quality content should naturally rise to the top of search results.

A close-up of a mouse cursor clicking on a box marked 'SEO'.

Trying to keep up with SEO best practices is a constant game of cat-and-mouse for anyone producing content online. (Image credit: Pexels | pixelcreatures)

Google isn’t entirely to blame here, of course. The study did note that Google performed better than Bing and DuckDuckGo in several key areas, and when we reached out for comment, a Google spokesperson concurred that “the study itself points out that Google has improved over the past year and is performing better than other search engines”. They also claimed that Google has “launched specific improvements to address these issues”, and highlighted that other third-party studies have found that “Google consistently surfaces higher quality results”, including this New York Times study from 2022.

However, there may be deeper problems at play than mere SEO spam attempts. The advent of generative AI tools such as the immensely popular ChatGPT has seen a marked increase in low-quality content specifically targeted to Google’s algorithm, and we’re only at the start of it.

The machines are coming… for your search results

The German team behind the study noted “a trend toward simplified, repetitive, and potentially AI-generated content”, and Bing and Google themselves are already introducing AI-produced search results, via the ChatGPT-powered Bing AI and Google’s own Google Bard. After all, why bother writing content and trying to optimize it for a search engine if a machine can do it for you?

There’s been a veritable avalanche of AI-generated content online since the dawn of ChatGPT, with many turning to machine learning tools to try to make a quick buck off the back of Google search traffic. Sites like the aptly named Robots.net are designed to churn out articles that target popular search terms, trying to draw traffic away from legitimate reviewers and journalists. These articles are assembled from the actual work of product reviewers and buying guide editors, scraped from dozens of reputable sites, and smashed together into something the AI thinks will please Google. Algorithms trying to satisfy algorithms… it’s a weird world, isn’t it?

You’ll notice I haven’t linked Robots.net despite our usual practice of linking directly to other news sites when we reference them, which is - somewhat amusingly - for SEO reasons. I don’t particularly want Google’s all-seeing algorithm to think we here at TechRadar are in any way associated with that sort of AI-authored drivel, but you can check it out for yourself if you’d like; it’s dismal stuff.

A screenshot of the Robots.net homepage.

Spoiler: the 'journalists' credited on AI-powered shovel-news sites like Robots.net don't actually exist. (Image credit: Robots.net)

As the reach of AI expands, the researchers warn that many internet users have already noticed “a torrent of low-quality content” that “keeps drowning any kind of useful information in search results.” I can’t say I disagree, and AI programs arguably have far greater capacity than human writers and editors to ‘game the system’ when it comes to SEO, so there’s a definite danger things are only going to get worse.

Of course, Google isn’t going to take this lying down. Combating AI spam is a major priority for the tech giant right now, although some have questioned Google’s sincerity on the matter - understandable, given Google’s own forays into AI use for search results. But I’d like to believe the powers that be at Google know that people don’t want content written by bots, for bots. Hopefully, the near future will hold changes that ensure the articles you get when you search for ‘best budget gaming PC’ are written by humans, for humans.

You might also like

https://ift.tt/KQfDIpy

No comments:

Post a Comment