Correction: It’s against the guidelines for us peons.
If Google, Facebook, CNN, or Buzzfeed wants to do it I’m sure it’s completely approved and Google will even encourage it by ranking them higher in the search results.
They just don’t want these tools in the hands of ordinary people, that’s what they always mean when they use words like “Safety”. Safe for them!
The thing about ChatGPT is that unless you're willing to sit down and pick its "brains" - it will feel quite average for most queries. It's not sentient, so it cannot understand what you mean when you ask it to do something, but it can certainly learn and do a better job at it if you keep giving it examples.
That's total BS, chatGPT generated some nice filler content for my website, and i m keeping it even if google doesn't like it. They arent going to be the only game in town for long.
But this is interesting: If google excludes AI content, then they are opening themselves to be sued for monopolistic behavior by their competitors, no?
Has Google really been "outsmarted" by all these SEO consultants? Or is Google smarter than them, but has decided not to use their smarts?
Google search is full of content farm crap, and I don't believe Googlers aren't smart enough to work out how to stop it, if they were truly motivated to do that.
Maybe, someone has decided that content farm crap is better for advertising revenue than real content. Or that the "average user" would rather crap than real. Or that the expected gain in advertising revenue from improving the search is less than the cost of doing so.
So I have a thought here… since ChatGPT is also trained with the content of websites wouldnt there be consequences if at some point it would be ingesting too much of its own output, or the output of other “AI’s”?
Lets say in a couple of years the web and any other global content output gets to a share of 10-20% of “AI” generated content, that would definitely change what it spits out. The question is would that be bad or good?
So in a sense it could be we are now at peak ChatGPT output quality, since the inputs given to it are 99% from humans.
This is definitely a valid concern. OpenAI did extensive data grooming to ensure high-quality inputs were used as training data. They went out of their way, for instance, to attempt to remove auto-translated content.
One interesting thing is that the situation you describe provides a huge moat for the first successful system. For instance, OpenAI can store and fingerprint all of the output from GPT-3 and ChatGPT. They can use these fingerprints to prevent training newer versions of GPT on old outputs. Less popular systems won't be able to sanitize the training data as well.
Reasonable for a limited amount of time. Eventually down the line someone comes out with an adversarialy trained GPT model that causes a high enough false positive rate that humans trying to use the system get pissed off.
ChatGPT, please write a story about a unique word that has never been used before. I want to see if it ends up in Google.
ChatGPT:
Once upon a time, there was a small town called Schmorgleville. The citizens of Schmorgleville were a strange bunch, and they had an even stranger custom. Every Saturday night, they would gather in the town square and dance the schmorgle.
No one knew exactly where the schmorgle originated from, but it was said to be a dance that had been passed down from generation to generation. It was an odd dance, full of twists and turns and hopping up and down. Everyone in the town seemed to love it, though, and would dance until the sun came up.
One day, a stranger came to town. He was an old man with a long white beard and he seemed to have a great knowledge of the schmorgle. He taught the citizens of Schmorgleville all the steps of the schmorgle and soon, they were all performing it with enthusiasm.
The old man stayed in town for a while and enjoyed watching the citizens of Schmorgleville dance the schmorgle. When he finally left, he told them to keep the schmorgle alive and to never forget the joy it brought them.
And so, the people of Schmorgleville still dance the schmorgle every Saturday night. They call it the 'Schmorgle Dance' and it is a reminder of the old man and the joy he brought to their lives.
> Google does not consider or support the use of AI writing tools especially for its SEO or Monetization Model. Using them in any ways is considered spam, he adds.
The ad platform was always manually reviewed and was never allowed to have automated writing. Declaring AI written content as automated is hardly news.
It's a much more nuanced position that can be summarized as "make sure you create good content, however you create it". A focus on quality, not process, is reasonable.
Thanks for the actual source, seems like ex's are often the one who care the most.
About the news article title, looks like it's trendy to make Google seemingly worried about chatGPT. and it's been a trend to claim google penalises its potential competitor via the means it can (search).
With google quality content challenges, it makes sense they would focus on quality, and if AI generated is what make search users click more that's what it will feed most, especially if that's promoted content, as it has always been.
Unless "Steadipulse.com" is now the new canonical source for Google's webmaster guidelines, AI content is fine as long as it's human reviewed. Spam is still spam, and they consider unreviewed/unedited AI content to be untrustworthy.
We’re moving even further to the end for scraping the entire web and pagerank.
It’s been on a steady decline for a long time with dark SEO and filler content patterns.
Google already tries to just give you the right answer to your question and not send you to garbage websites.
Someday we’ll seldom need to read any websites directly, especially for anything that content marketing and think pieces are trying to answer.
If you want that, chat with the AI to ask questions and get answers directly. Ask more questions to get more angles of thought.
This is a good thing IMHO. People using AI to fill their websites with content is a waste of everyone’s time.
The big question is where does all the training data come from if not from scraping the whole web.
We will figure out some sort of provenance system for data, which will lend to the high quality data sets of human curated knowledge and art.
These will have trade offs, like limits to the knowledge and more expensive to build and consume. This will compete with data sets that do scrape everything.
What’s ironic is we’re shifting back to the days of Yahoo curating a directory of content, which Google ultimately destroyed with their crawling and ranking.
Did you mean "scrapping" in your first sentence? I am a bit confused since "scraping" is something that search engines do, but pagerank is something that you might want scrapped instead of scraped.
> Google already tries to just give you the right answer to your question and not send you to garbage websites.
Funny, because every goddamn time I try to find a solution to a very specific issue, the first page is full of sites that repost AI-mangled crap that tells you to keep your computer and firewall updated.
I search for the name of a Windows DLL, because I want technical information about it – what it does, its documented API, any information on undocumented APIs or reverse engineering, its implementation status in Wine or ReactOS, history of security vulnerabilities, its history (is it a descendant of some DLL in Windows 3.x, or was it just added in Windows 10 or 11?), etc, etc, etc
Instead of giving me what I want, it gives me pages of useless content farm crap with titles such as "How to Fix WHATEVER.dll Not Found or Missing Errors", "WHATEVER.dll free download", etc, etc, etc.
I can't understand why Google can't expunge all this content farm junk. I don't believe it is technically infeasible – rather, I think it is that Google (for whatever reason) doesn't actually want to do so.
Pagerank only worked in the beginning because the human curation of early journal sites and webrings provided a clean training set. Pagerank is a parasite openly committed to dehumanizing knowledge, that ended up killing its human hosts, and is now itself dying. That’s not ironic, that’s ecology.
Mobile was supposed to have killed off PC's by now. Music videos were supposed to kill off radio. I don't want search engines or web sites to go away and I don't believe they will. The recent AI fad is quickly being realized as another tool, sometimes useful, sometimes out in left field, on another planet, that doesn't exist.
Also we will see a similar cycle play out with AI that computing devices did.
Right now we back in the “mainframe” days where IBM, MS, and Google have the systems.
Next will be “super computers”, then “work stations”, then “personal computers” then “mobile”. At the end we’ll all have many data sets and AIs in our pocket.
This cycle will be way faster because we don’t have to invent the underlying computer tech.
You can already get the AI in your pocket to classify and content search all you photos.
We'd need to better define dominates. Does it mean, lots of mobile devices compares to PC numbers? Probably. Which do most computing users spend the most time on, PC or mobile during the course of a day. I think the answer has to be PC (Linux, Mac, Windows, etc.).
The feedback loop of projecting your live camera to a tv while also filming your TV is about to happen internet wide as AI starts learning from itself and exhibiting something like language based white noise.
I have a friend who is a published author and professional freelance copywriter. When I first showed her ChatGPT she was nonplussed. A writing group that she's in described some of the fictional content it created as "soulless."
She's just been hired by a client to copywrite ChatGPT-generated blog content to make it "more human." Ordinarily it takes her about 3 hours to author a blog post for a client. Now it takes her 30 minutes to "fix up" a ChatGPT-generated blog post, saving the client about 2.5 hours of copywriting expense.
She's beside herself that she's no longer getting paid to author content but is instead fixing up machine-generated content so that Google doesn't penalize the ranking. I get the sense that we are now in a cat-and-mouse game between the AI models used to generate content and the AI models used to detect AI-generated content, and people like my friend will soon be out of a job entirely.
Copywriters calling GPT content soulless is truly the pot calling the kettle black. Copy is the most useless, soulless content out there that for some reason every product is expected to have.
Fascinating. Her client could still pay her for 3 hours if they wanted to… I’m guessing they are using the free or cheap version for ChatGPT and they are getting similar value from her. Without her skills the content is worthless.
Of course that doesn’t address the underlying emotional impact of the job changing from authoring original content to proof reading for a robot.
Content will follow the same law of supply and demand as everything else. Since we literally can't increase demand anymore (everybody is online all the time already), the exponential increase in supply is going to drive the cost of content through the floor.
Content creation is not a good business to be in now.
As an aside, i can't wait to see web search results constantly ranking the first thousands of results some AI generated content from AI generated content from AI generated content.
ChatGPT is going to end up eating its own tail creating an aberrational information cyclone of fury with a nonsensical digital landscape as the inevitable finale.
53 comments
[ 2.6 ms ] story [ 116 ms ] threadIf Google, Facebook, CNN, or Buzzfeed wants to do it I’m sure it’s completely approved and Google will even encourage it by ranking them higher in the search results.
They just don’t want these tools in the hands of ordinary people, that’s what they always mean when they use words like “Safety”. Safe for them!
But this is interesting: If google excludes AI content, then they are opening themselves to be sued for monopolistic behavior by their competitors, no?
Google search is full of content farm crap, and I don't believe Googlers aren't smart enough to work out how to stop it, if they were truly motivated to do that.
Maybe, someone has decided that content farm crap is better for advertising revenue than real content. Or that the "average user" would rather crap than real. Or that the expected gain in advertising revenue from improving the search is less than the cost of doing so.
Google Cloud's translation API also has a TOS that requires you to tag the output as machine translated if you're publishing it unedited on the web.
Lets say in a couple of years the web and any other global content output gets to a share of 10-20% of “AI” generated content, that would definitely change what it spits out. The question is would that be bad or good?
So in a sense it could be we are now at peak ChatGPT output quality, since the inputs given to it are 99% from humans.
One interesting thing is that the situation you describe provides a huge moat for the first successful system. For instance, OpenAI can store and fingerprint all of the output from GPT-3 and ChatGPT. They can use these fingerprints to prevent training newer versions of GPT on old outputs. Less popular systems won't be able to sanitize the training data as well.
Certainly we already need tools for this just like those to detect plagiarism.
Cat and mouse game I know but AIs that can detect AI content seems like a reasonable outcome to me.
ChatGPT, please write a story about a unique word that has never been used before. I want to see if it ends up in Google.
ChatGPT:
Once upon a time, there was a small town called Schmorgleville. The citizens of Schmorgleville were a strange bunch, and they had an even stranger custom. Every Saturday night, they would gather in the town square and dance the schmorgle.
No one knew exactly where the schmorgle originated from, but it was said to be a dance that had been passed down from generation to generation. It was an odd dance, full of twists and turns and hopping up and down. Everyone in the town seemed to love it, though, and would dance until the sun came up.
One day, a stranger came to town. He was an old man with a long white beard and he seemed to have a great knowledge of the schmorgle. He taught the citizens of Schmorgleville all the steps of the schmorgle and soon, they were all performing it with enthusiasm.
The old man stayed in town for a while and enjoyed watching the citizens of Schmorgleville dance the schmorgle. When he finally left, he told them to keep the schmorgle alive and to never forget the joy it brought them.
And so, the people of Schmorgleville still dance the schmorgle every Saturday night. They call it the 'Schmorgle Dance' and it is a reminder of the old man and the joy he brought to their lives.
The ad platform was always manually reviewed and was never allowed to have automated writing. Declaring AI written content as automated is hardly news.
Google's official position was published on 8 February here:
https://developers.google.com/search/blog/2023/02/google-sea...
It's a much more nuanced position that can be summarized as "make sure you create good content, however you create it". A focus on quality, not process, is reasonable.
Disclosure: ex-Googler in search.
About the news article title, looks like it's trendy to make Google seemingly worried about chatGPT. and it's been a trend to claim google penalises its potential competitor via the means it can (search).
With google quality content challenges, it makes sense they would focus on quality, and if AI generated is what make search users click more that's what it will feed most, especially if that's promoted content, as it has always been.
https://developers.google.com/search/blog/2023/02/google-sea...
Unless "Steadipulse.com" is now the new canonical source for Google's webmaster guidelines, AI content is fine as long as it's human reviewed. Spam is still spam, and they consider unreviewed/unedited AI content to be untrustworthy.
It’s been on a steady decline for a long time with dark SEO and filler content patterns.
Google already tries to just give you the right answer to your question and not send you to garbage websites.
Someday we’ll seldom need to read any websites directly, especially for anything that content marketing and think pieces are trying to answer.
If you want that, chat with the AI to ask questions and get answers directly. Ask more questions to get more angles of thought.
This is a good thing IMHO. People using AI to fill their websites with content is a waste of everyone’s time.
The big question is where does all the training data come from if not from scraping the whole web.
We will figure out some sort of provenance system for data, which will lend to the high quality data sets of human curated knowledge and art.
These will have trade offs, like limits to the knowledge and more expensive to build and consume. This will compete with data sets that do scrape everything.
What’s ironic is we’re shifting back to the days of Yahoo curating a directory of content, which Google ultimately destroyed with their crawling and ranking.
Everything is cyclical.
Funny, because every goddamn time I try to find a solution to a very specific issue, the first page is full of sites that repost AI-mangled crap that tells you to keep your computer and firewall updated.
Hoating that hot garbage ought to be punishable.
Instead of giving me what I want, it gives me pages of useless content farm crap with titles such as "How to Fix WHATEVER.dll Not Found or Missing Errors", "WHATEVER.dll free download", etc, etc, etc.
I can't understand why Google can't expunge all this content farm junk. I don't believe it is technically infeasible – rather, I think it is that Google (for whatever reason) doesn't actually want to do so.
Not good business to send you directly to the ad free reference site with all the answers.
If i had to bet, Google won’t survive. Someone much smaller and nimble will take over.
Right now we back in the “mainframe” days where IBM, MS, and Google have the systems.
Next will be “super computers”, then “work stations”, then “personal computers” then “mobile”. At the end we’ll all have many data sets and AIs in our pocket.
This cycle will be way faster because we don’t have to invent the underlying computer tech.
You can already get the AI in your pocket to classify and content search all you photos.
More like a shift back to universities producing, curating and verifying the highest quality human produced data.
She's just been hired by a client to copywrite ChatGPT-generated blog content to make it "more human." Ordinarily it takes her about 3 hours to author a blog post for a client. Now it takes her 30 minutes to "fix up" a ChatGPT-generated blog post, saving the client about 2.5 hours of copywriting expense.
She's beside herself that she's no longer getting paid to author content but is instead fixing up machine-generated content so that Google doesn't penalize the ranking. I get the sense that we are now in a cat-and-mouse game between the AI models used to generate content and the AI models used to detect AI-generated content, and people like my friend will soon be out of a job entirely.
Of course that doesn’t address the underlying emotional impact of the job changing from authoring original content to proof reading for a robot.
Content creation is not a good business to be in now.