Why Is Google AI Telling You To Eat Glue? AI Gemini Overviews And The Memes About Them Explained

It used to be that you Googled something and you received several hundred thousand results from around the internet in a fraction of the second. Now, when you Google something, you received two sentences generated by an AI telling you to eat glue or rocks. The internet of 2024 is not what it used to be, and that change is probably most clearly written on the walls of the internet's front foyer: Google Search.

What's happening with the new Google AI Overview product, and why does essentially everyone hate it? Let's explain.

What Is Google Gemini AI Overview?

Google Gemini AI Overview is a new feature added to Search, Google's flagship product, which generates an answer to a user question using AI. Hypothetically, this is supposed to save a user's time and maximize productivity. In practice, however, users have noticed that the AI overview lies, hallucinates and says unhinged stuff at an alarming rate.

Why Did Google Create An AI Overview Feature?

Google claims that the AI overview feature will make the experience of search more seamless for users. Instead of having to choose an answer yourself from a series of listed results, the AI will parse the information for you and serve it up in a rapid, digestible format.

Why Is The AI Overview So Bad?

There are a few theories about why the Google AI Overview feature is so ineffective. Some think it may have been a rushed rollout by the company in an attempt to keep up with rivals in the AI and Search fields such as Microsoft's Copilot and OpenAI. Others think it's an extension of problems that have previously plagued Google AI products like the Gemini image generator.

Another possible explanation of the problem could be Google's use of data from Reddit and the inability of artificial intelligences to distinguish sh*tposts from factual information. By this theory, the AI might be scooping up jokes and trolls by Redditors and interpreting them as truth. Another source of misinformation shared by the AI might be satire and ironic websites such as The Onion.

Is This Proof Of Dead Internet Theory?

The Dead Internet Theory proposes that the majority of activity on the modern web is generated by bots rather than humans. If the Dead Internet Theory is correct, then the deep currents that move internet culture and tech company decision-making are not the market pressure brought by human beings wanting stuff, but the programmed automatic behaviors of bots bouncing off of and responding to other bots.

Google appears to be choosing the AI overview because it optimizes users' time spent on the platform, but it's hard to know how much of what Google measures is actually human beings. Further, a lot of what Google's AI is responding to and ranking in SEO results and AI data training sets is text written by other AIs. So, in a certain sense, Google is eating itself alive. AI is imitating bots which are imitating other bots which are imitating humans who (because of how social media works) are maybe imitating bots themselves in order to get good rankings on SEO and see their posts spread far and wide.

The whole thing is far too tangled and confusing but somehow it is still profitable, perhaps because Google controls nearly 90% of the search market according to the latest estimates and is, essentially, a monopoly.

For the full history of Google AI Overviews, be sure to check out Know Your Meme's encyclopedia entry for more information.