Google now won't black-hole all AI-made pages as spam

Web content automatically generated by AI will be ranked according to its quality by Google Search, the internet titan confirmed this week, after previously suggesting it would blanket down-rank computer-created pages. Last year, Google Search Advocate John Mueller said using machine-learning tools to produce text is " …

[Howard Sway]

Will these models actually be useful and we can all spend less time browsing the internet?

We'll all spend less time browsing the internet using these search engines.

Once a few people have got these models churning out billions of garbage AI generated webpages a day, and the search engines have then crawled this garbage and added it to their web of "information", they will essentially become useless. Everyone will get furious and someone else will have to find a way of supplying reliable information that avoids all meaningless AI "content".

[CatWithChainsaw]

Re: Will these models actually be useful and we can all spend less time browsing the internet?

Hopefully high schools still have those giant World Book Encyclopedias on the shelves!

[EricB123]

Re: Will these models actually be useful and we can all spend less time browsing the internet?

I had World Book encyclopedias when I was in elementary school. I don't recall those to be terribly accurate either.

[Anonymous Coward]

Re: Will these models actually be useful and we can all spend less time browsing the internet?

Encyclopedias (and Wikipedia these days) are virtually always 100% accurate when they are created, science and world researchers may find issues over time and the encyclopedias were almost always rewritten when the changes were seen as accurate, for example the original idea that Jewish slaves built the pyramids was probably wrong, but that was happening thousands of years before the first post on Twitter.

[teknopaul]

Re: Will these models actually be useful and we can all spend less time browsing the internet?

Google ceased being useful a while back, it is all spam now. Often pages of adverts before any actual content.

"It's important to recognize that not all use of automation, including AI generation, is spam,"

Means "Thats what Google search has become, so we can't say bad things about it."

AI for Google is simply a way to crawl the Internet and spit it back out without quoting sources or providing links. It's copyright infringement of the whole Internet in one go. Much like amp and all its previous attempts to do the same.

But this will fail imho, because content without quoted sources has become worse than useless. Likes of Trump have been gaming that system from before Google decided to bank its future on it.

No wonder their stock tanked.

Search engines need to search the Internet and return results with the quoted source first and foremost: which is what they did back when they were useful.

Use qwant.

[captain veg]

Web content automatically generated by AI will be ranked according to its quality

"Web content automatically generated by AI will be ranked according to its quality by Google Search, the internet titan confirmed this week"

In which case, we've nothing to worry about. It's all shit.

-A.

[yetanotheraoc]

Re: Web content automatically generated by AI will be ranked according to its quality

"It's all shit."

Google's own AI generated content will be the good shit.

[werdsmith]

Re: Web content automatically generated by AI will be ranked according to its quality

I’ve tried looking up the word “celui” on google translate, French to English and I’m a bit bemused by the result. I wonder if others get the same? Presumably an AI aberration.

[Anonymous Coward]

Re: Web content automatically generated by AI will be ranked according to its quality

it's the same in a few languages, correct in others... bizarre.

[Version 1.0]

Re: Web content automatically generated by AI will be ranked according to its quality

Is AI thinking that people are snorting Coke, or drinking Coke, but never trying Pepsi?

[Neil Barnes]

Re: Web content automatically generated by AI will be ranked according to its quality

They could hardly do otherwise; to ban AI generated pages would have been to put themselves in the embarrassing position of being unable to show their own home page.

FFS guys: if I use a search engine, it's because I'm looking for something, not for a machine generated statistical blurb about what the something might be. Gimme sources, not bullshit.

[M.V. Lipvig]

Re: Web content automatically generated by AI will be ranked according to its quality

Found - 284,836 links to bullshit sources.

Betcha "bulls" will be 30 pages in at least.

[Kevin McMurtrie]

Justifying the junk

Google search results for product reviews and tech support are worthless because the results are bot garbage. The way I read this, Google has given up on filtering them out going forward.

[FF22]

Google BS

For one, Google had this policy for at least a year if not two now. Also, they have this policy, because they're unable to tell AI generated content from human-authored content - so, they don't even try to bother.

That being said, Gogole also can't judge quality of content either, and they have been ranking low-quality AI-generated, machine translated content above high-quality human generated for years now. Any human looking at just the search result list could tell most AI generated content just by looking at the nonsensical domain names they're using or by the excerpt they're showing - yet Google itself is unable to do even basic filtering on these.

Unfortunately all this will (continue) lead to original, human content creators to stop generating content, and everything you will find through Google will regurgitated content re-written over and over by an AI, that will feed more and more on other BS content generated by other AIs.

AI will be the demise of Google, but not because ChatGPT will kill classic search, but because search itself (be it the classic form or that through ChatGPT) will become pointless and only serve up low-quality and unreliable content.

Now, the obvious solution would be to rank pages based on competence and such - but that would require an actual human understanding and evaluating the content, which, however, is something that's Google is terribly afraid of, because it costs actual money and would mean that they also have to invest, not only reap the benefits of content created by millions of others.

[yetanotheraoc]

$120 bn

"Over $120 billion of Google's market cap was wiped shortly after the error."

That's catchy, but Alphabet is still trading 5% higher than it was a month ago.

[xanadu42]

Oh for the days...

When a search engine would provide matches ranked on all the words (or phrases) you searched for and then start whittling down based on less and less matches...

Now search results show what Google (et al) *THINK* you are searching for and you have to wade through so much crap...

AI (more correctly ML) is going to make finding what you are actually searching for even more arduous

[Anonymous Coward]

Re: Oh for the days...

Exactly this - Google has consistently got worse over time at turning up relevant results for some search terms I have an interest in. Pages that I know still exist don't even make the results any more, never mind being on pages 2, 3, 4, etc.

[nintendoeats]

Re: Oh for the days...

I am reminded of a story from some Microsoft corpo, over 10 years ago. May have been Steve Ballmer, but I don't remember for sure.

He was talking about how he was using a hot air hand dryer once, and "realized it was the smartest device I had ever used" or some such. He waxed lyrical about how it knew exactly what to do at all times, with basically no human interface. He wanted their devices to be "smarter", like the hand dryer.

Problem with this reasoning: the hand dryer isn't smart. It's dumb. It has one user input (range sensor), one output (hot air), and the relationship between the two is trivial to understand. And THAT is why it's easy to use; because it's so freaking predictable. If it were smarter, it would be harder to predict and possibly have a worse UI.

Make things as smart as they need to be and no smarter. If the user can no longer construct a mental model of what inputs will give the desired outputs, in my opinion the UI is broken.

[breakfast]

There may be a reason he thought it was smart

Ironically Steve Balmer also had one output, and that was also hot air.

[Filippo]

Bing can't answer either

The reason a wrong answer by Google's model caused Alphabet to drop 5% is that search is its core business. Any weakness there is cause for serious concern. On the other hand, nobody uses Bing anyway, so even if its bot is just as wrong as Google's, it's as much of a problem for Microsoft.

If someone else integrated a LLM into an operating system (*shudder*), and Microsoft then tried and failed to do the same, now that would put a big dent in MSFT's cap.

[Anonymous Coward]

Re: Bing can't answer either

I beg to differ, Bing is much better for torrents and porn.

[Anonymous Coward]

Re: Bing can't answer either

A shoutout to my fellow Bing users. We do exist!

[Ken Moorhouse]

E-E-A-T – expertise, experience, authoritativeness, and trustworthiness

I can't seem to extinguish from my brain the similarity this has with something else.

[Ken Moorhouse]

Mood Swings

With AI, when I search for something, and it comes back with ridiculous results, and I refine my search, will the system come back and say:-

"Look, I've already told you what you want to know, are you disputing my authority on this subject?"

[that one in the corner]

Re: Mood Swings

Respect mah authoritaaaay

[Binraider]

So, tools in their infancy get to spread misinformation that is probably going to be wrong; and it's not filtered out by the worlds most used search engine.

This won't end well.

For developers of alternate search engines (DuckDuckGo comes to mind); if you really want a slice of the pie, here's the opportunity to differentiate yourselves by filtering out AI generated trash.

[a pressbutton]

So, at some point there will be an LLM (A) posting content with comments switched on and another LLM (B) will use Google to find that content and will post a comment and ...

... A and B will start to talk to each other

[jpennycook]

AI and newspapers

I wouldn't be surprised if many stories on local newspaper websites were AI-generated (or maybe lightly edited after AI), given that so many seem to be duplicating press releases, Trip Advisor reviews, or copying from council meeting minutes. No doubt this will be the case in future.

[Hubert Cumberdale]

Re: AI and newspapers

If my local rag is anything to go by, an AI could probably do better than the humans that write for it most of the time.

[Anonymous Coward]

Another link in the chain

It is already tedious to find the original article among lots of summarizing copycats "as reported by (Other Web Site)".

I'd like to read the source, not someone's commentary to watching someone else read it.

[T. F. M. Reader]

Quality assessment

If it looks like it was written by the Bard it will be judged acceptable?

[T. F. M. Reader]

Reducing the number of eyeballs on pages

Thus saving advertisers gazillions? And maybe killing Google's business model at letting MSFT win? Or will the Bard engage in product placement now?

[M.V. Lipvig]

Re: Reducing the number of eyeballs on pages

I can see it now.

Search - latest Ukranian war news.

Results - Ukrainian warriors swear by ice cold Coca Cola during red hot firefights.

- Russians seen retreating comfortably in Nike running shoes.

- Russia tells world that war will continue until Kentucky Fried Chicken returns to Moscow - it's that good!

- Ukraine says Russia is after their supply of Tide pods

- Russia and Ukraine agree on one thing - Costco chicken!

Giogle.uk may return slightly different results.