You’re trying to cook pizza from scratch and you just can’t figure out why the cheese keeps sliding off. You agitatedly pull up Google and look up “cheese not sticking to pizza” in hopes of finding a solution. Luckily, the helpful Google AI Overview will dig through the search results and present some solutions directly to you! You’ve got some useful tips now; “Mixing cheese into the sauce helps add moisture to the cheese and dry out the sauce.” Weird, but okay. “You can also add about ⅛ cup of non-toxic glue to the sauce to give it more tackiness.” Alright, less okay.
The instance described above was a real piece of advice proposed by Google AI Overview, as discussed in a Wired article from earlier this year. As funny as it sounds, in reality, it’s a frustrating, disturbing example of how unreliable Google is becoming due to its over-reliance on artificial intelligence.
Imagine how easily AI could destroy your credibility in an essay, nullify your argument in a debate, or unwillingly involve you in the spread of misinformation when you put your trust in it.
Google’s self-proclaimed “AI Perspective” says that “AI can unlock new scientific discoveries and opportunities and help tackle humanity’s greatest challenges—today and in the future.” Tragically, eating glue probably doesn’t align with these lofty ideas of enabling scientific discovery.
Yet another worrying factor of AI overviews like Google’s is their process. They basically skim the search results of your topic and pull what seems like relevant information before packaging it up into a simple summary. These summaries can have details lost in communication or can pull information from unreliable or satirical sources before presenting them as fact.
These instances are called “hallucinations,” or when AI will present false information. Fortunately, Gemini, the AI that powers Google’s overview, recognized that it is not infallible and presented a concise, thoughtful solution! Their FAQ recommends that you should “basically double-check Gemini’s responses.”
Their recommendation is to Google your Google. At that point, why not just start with reading a human-written article? Why isn’t Google just starting by boosting those credible sources to the top, rather than presenting misinformation? Why are we being told to eat glue?
There are a couple of alternatives to swerving the omnipresent Google AI. For one, you can manually use the filters Google offers. For instance, jumping directly to the “news” tab after hitting enter will lead to a list of articles rather than the AI response.
Further, you may install a browser extension that automatically directs you to AI-free search results. Finally, you can always use a different browser altogether. DuckDuckGo is my favorite. DuckDuckGo does have its own AI assistant, DuckAssist, which is currently in beta testing. But unlike Google, it asks if you would like an answer to be generated for you. It won’t automatically formulate one or tell you unprovoked that you should eat glue.
AI is a prominent feature in everyday life now, and it’s only going to become more so as we keep tripping over our feet into the future. Awareness of the fact that AI can be (and often is) unreliable has become an absolutely necessary tool.