If You Ask Google Why It Sucks Now, AI Overviews Will Viciously Bully Google and Itself

Google’s notorious AI Overviews feature appears to have finally gotten it right — but for all the wrong reasons.

As flagged on Bluesky, if you ask Google why its own search engine feels so useless lately, the company’s AI feature will provide multiple and excoriating reasons — including brutal attacks against itself, Google’s much-maligned AI Overviews.

In that Bluesky post, Google responded broadly to the hilariously colloquial query of “why is google ass now” with a self-flagellating list of bullet points. The AI’s detailed rejoinder included charges that the search giant has been accused of prioritizing “AI-generated and SEO-optimized content” and that its “AI-generated answers” — yes, provided by the very entity answering the question — have sparked “accuracy concerns,” including “instances where Google’s AI provided harmful or nonsensical suggestions” as well as claims that are “inaccurate or misleading.”

Inspired by the vulgarity, we asked Google why it’s now “so f*cked up” — and got a similarly self-deprecating explanation that again demonstrated that the bot shows a striking lack of loyalty to its creator, or regard for its own reputation.

In its response to our query, the AI pointed out that the search engine is now “cluttered” with “ads and self-promotion.” It also noted that Google has been criticized for “manipulation of search rankings” and ingesting “AI-generated content” — the same kind of comical self-criticism we’ve seen from various AI models time and again.

“The rise of AI-generated content can lead to low-quality, keyword-stuffed articles that rank highly but offer little genuine insight,” AI Overviews said, adding that “Google’s AI Overviews have also been criticized for providing inconsistent or overly simplistic advice based on questionable sources.”

Curious about how Google might respond when asked a bit more politely about its recent losing streak, we dropped the obscenity and queried “why is google useless.” Again, the response pointed to the same common critiques and flaws.

“Google’s search engine is often criticized for being less effective at finding specific or complex information, with some users finding results cluttered with ads and irrelevant content,” the response read. “This perception is influenced by… Google’s shift towards AI-powered summaries that may not always be accurate or comprehensive.”

It also offers suggestions for alternatives, pointing users to the competing search engines offered by Microsoft’s Bing and DuckDuckGo.

AI Overviews has also been criticized for eviscerating traffic to news publications, by letting readers get the answers they’re looking for without actually visiting — and thus supporting, via ads or subscriptions — the sites that actually did the work of providing the information.

Spokespeople for the tech giant have rationalized that AI Overviews does offer citations, though research has found that the clickthrough rate on these links is abysmal, threatening what vestiges remain of the news media.

But the citations themselves are also often bizarre and random, often seeming to prioritize help forums, social media, non-authoritative blogs and sketchy publications.

Ironically, this was abundantly clear in the sources cited by the AI for why Google has become so useless. For instance, the AI cited a post from Google Help from 2022, several out-of-date posts on Reddit, a Quora post from 2021, and a junky SEO-optimized blog post from last year.

The only particularly trustworthy-feeling sources from the “useless” query, in fact, were a short explainer from a University of Iowa librarian posted in April — which seemed to be aimed at nursing students and not the general public — and a Mashable article about AI Overviews itself from May of this year that found that the AI still “struggles to answer basic questions.”

In other words, even when criticizing its own creator, AI Overviews is citing some seriously dubious material. The user-generated posts from 2021 and 2022 are particularly immaterial because they reflect an entirely different era of Google’s so-called “ensh*ttification” — one that seemed egregious at the time, but seems quaint in hindsight now that search engine — and, to be fair, the web itself — are being eaten by AI.

There could be plenty of reasons why Google’s AI is glomming onto such low-grade material. The company’s long-running algorithms have, for instance, often been criticized for their opacity, especially when it comes to the ways they rank prestigious outfits like the New York Times and CNN over smaller and more localized outlets.

With such mysterious inner workings, there now appears to be a similar problem going on with AI Overviews — except that now older, low-quality, and user-generated content appears to be suppressing authoritative and mainstream sources instead of the other way around.

Add in the tech world’s overarching obsession with stuffing generative AI — the same models that regularly make stuff up and can’t tell dates or time — down everyone’s throat, and you get a perfect storm of AI slop right at the top of Google’s original and flagship product: its eponymous search engine.

In the two years since Google announced that it was injecting AI into its search results, we’ve seen the feature — which used to be called “Search Generative Experience” before rebranding to “AI Overviews” in May 2024 — go from suggesting users eat rocks and add glue to their pizza to making up nonexistent etymologies for weird and fake idioms like “you can’t lick a badger twice” or “the bicycle eats first.”

We’ve reached out to Google to ask if it has anything to say about AI Overviews criticizing the company and itself, and we’ll update if we hear back.

More on Google: Google Cofounder Exhorts Employees to Work 60-Hour Weeks to Create AI Designed to Replace Them

The post If You Ask Google Why It Sucks Now, AI Overviews Will Viciously Bully Google and Itself appeared first on Futurism.