OPINION: Google I/O has historically been an event used to showcase the latest updates coming to Android. However, this week, Google chose to focus on AI instead, and specifically how it will change how we navigate the web.
Google’s search engine is arguably the most important asset that the company owns, acting as the launchpad to make Google the technology giant it is today. According to Statista, Google dominates a whopping 82% of the desktop search engine market share, as of January 2024. In second place is Bing, with just 11% of the share.
And yet despite its dominance, Google has announced that it will be drastically overhauling its popular search engine, by integrating a new feature called AI Overview.
This new feature leverages advanced artificial intelligence technology to generate unique answers for many search queries. As the answers are generated by Google they will be embedded into the results of the search engine, meaning you won’t need to click on any link to actually find the information you’re looking for.
I can certainly understand the appeal of this for the average user, as you’re able to get all of the information you need more quickly, while also having it all neatly summarised so you don’t have to scroll through lines of text. Google even offers you the option to simplify the summary even further, or even break it down into more detail.
That all sounds great on paper, but I’ve got a lot of concerns about where Google is getting all this information from. Google does at least give credit to the websites and articles that it’s harvesting the information from, but given the nature of AI, it’s not always perfect at correctly summarising that information.
Google has also been known to take information from unreliable sources, and then present incorrect information as its ‘Featured Snippet’ in the search results. You can check out the video from Vox below, which showcases some of these significant mistakes, including a post that suggests dinosaurs never existed, and were simply a construct to indoctrinate children.
That video is admittedly seven years old now, and Google will argue that it’s improved its search results significantly in that time frame, but there are a few major mistakes of Google’s new AI Overview already going viral.
While I can’t verify this search result, one post on X (formerly Twitter) claims that Google’s AI Overview suggested that anyone suffering from kidney stones should drink urine every 24 hours to help them pass.
Having medical advice that is unvetted by a professional or expert is incredibly dangerous, and Google is running the risk of doing this by rolling out its new AI Overview feature. In fact, even beyond medical advice, Google is unable to ensure that all the information it’s harvesting is coming from an expert.
Google recently penned a deal with Reddit, allowing the tech company to use posts from countless forums to train its AI. During Google I/O, a demonstration video also showed Quora pop up as one of the sources of information used for the answer generated by AI Overview.
Forums are a valuable place to get information from other people across the web, but since anyone can post without any professional knowledge or fact-checking process, there is no way to guarantee that what you’re reading is correct. That’s completely fine if you’re aware of this when scrolling through answers in Reddit and Quora, but it becomes a lot more dangerous when the information is presented as being factual from a trusted source such as Google.
Of course, that’s not to say you wouldn’t still find incorrect information when using Google’s search engine and clicking on a link to visit a website. The difference here, though, is that the reader can make up their own decision as to how trustworthy the information is.
For example, you’re far more likely to think the information is accurate when reading it on the BBC as opposed to an amateur blog. But with Google, many people may incorrectly assume that Google is just as reliable as any prominent publication for providing factual information, which sadly isn’t the case.
With the world wide web already rife with misinformation, I’m growing very concerned that Google’s push into AI won’t actually solve this problem, but intensify it instead. An answer generated by AI is always going to be more prone to error than one written and fact-checked by an experienced journalist, and Google needs to understand that quickly before it damages the quality and credibility of its search engine.
Ctrl+Alt+Del is our weekly computing-focused opinion column where we delve deeper into the world of PCs, laptops, handhelds, peripherals and more. Get it straight into your email inbox every Saturday by signing up for the newsletter.