How is Google able to differentiate bw AI and Human content?

[googgle]

They can’t differentiate that. Unless ur AI is crappy.

 

[rifmedia]

That's just a sign of poor content, not AI content.

AI content can't solve problems, the most of question I personnaly ask Google are:
- Health question when I suffer by something ...
- Programming questions at work.
- Religion questions...
None of this questions can AI give me a good answer, so for me IA content is really a poor content.

 

[DanielX]

I don't understand. Why Google should penalize AI content at all? If the content is good, original, factual correct and the user loves it, staying long time and returns, imho the content should rank. If the content is crap, making the user to bounce immediately or copied it should be deindexed, no matter if its human or AI crap. After all, a search engine should be focused on quality and user experience and not on a ranking contest between seo gurus. This is not a chess competition where AI already outranks humans and has to be forbidden to enter the contest. Otherwise we will in future have Google dominated SERPS with human content, a lot of it crap, and another Search engine with AI/human mixed SERPS with additional quality content and less crap. I would use the latter. All the PAA questions I might have I already now ask ChatGPT first and rarely feel the need to consult Google afterwords.

 

[DanielX]

AI content can't solve problems, the most of question I personnaly ask Google are:
- Health question when I suffer by something ...
- Programming questions at work.
- Religion questions...
None of this questions can AI give me a good answer, so for me IA content is really a poor content.

Have you tried to submit your questions to ChatGPT?

 

[theodor_adornor]

I don't know otyher AI creators but I have read it that OpenAI researchers developed cryptographic watermarking which is created through an OpenAI product like ChatGPT. So, maybe Google follow those Watermarkings?

 

[Hunter112]

I don't understand. Why Google should penalize AI content at all? If the content is good, original, factual correct and the user loves it, staying long time and returns, imho the content should rank. If the content is crap, making the user to bounce immediately or copied it should be deindexed, no matter if its human or AI crap. After all, a search engine should be focused on quality and user experience and not on a ranking contest between seo gurus. This is not a chess competition where AI already outranks humans and has to be forbidden to enter the contest. Otherwise we will in future have Google dominated SERPS with human content, a lot of it crap, and another Search engine with AI/human mixed SERPS with additional quality content and less crap. I would use the latter. All the PAA questions I might have I already now ask ChatGPT first and rarely feel the need to consult Google afterwords.

because no one creates new information, and it takes Google too much to process the information

 

[sagarpatil]

The metric is how visitors interecat with this content, do they spend lot of time on the website, do they visit more than one page, do they come back and click on others results on search ...

This makes perfect sense from Google's point of view. They are here to make money, they don't really care who write the content. If AI is able to generate better content than humans and if the users find it useful, what exactly is the problem?

 

[sagarpatil]

I don't know otyher AI creators but I have read it that OpenAI researchers developed cryptographic watermarking which is created through an OpenAI product like ChatGPT. So, maybe Google follow those Watermarkings?

I don't think they've developed it, the last I heard, they just did some research on it. I'm not sure how developing this technology will help them make money. Even if they do, I doubt it will be open-sourced. Remember, Microsoft has invested in OpenAI. I think MSFT could probably use the tech to find AI content on Bing.

 

[bartosimpsonio]

Human reviewers and competitor snitching. Otherwise they can't tell.]

I'm 99% sure most Q&A sites are like 80% AI generated questions and replies, just to give one example. They rank like fire.

 

[danetf]

ai keeps improving and I think what google would be able to do is pass a law just like what china did. Or Encourage users to Add a watermark to an article or image if it's generated by ai. Ai would be better than human writing in the next 5years.

 

[regcleaner]

Google uses a combination of automated systems and human review to differentiate between AI-generated content and content created by humans. One way it can do this is by using natural language processing (NLP) algorithms to analyze the text and determine whether it reads like something that a human would write. Google may also use other signals, such as the source of the content and its overall quality, to help determine whether it was created by a human or by an AI system.

It is important to note that Google's ability to differentiate between AI-generated and human-generated content is not perfect, and it is possible for AI-generated content to be mistakenly identified as human-generated or vice versa. However, Google is constantly working to improve its systems and algorithms to better differentiate between AI and human content.

 

[blackbeans]

Here's what OpenAI is doing https://www.newscientist.com/articl...ermark-to-identify-work-from-its-gpt-text-ai/
Expect Google to get their hands on such markers

 

[Johnpeterburs]

Not everything can be differentiated, but more with Sentimental Analysis Metrics.

If a trained model is prepared to produce good manifested content, then the other AI model can be trained to detect the same. So it is highly possible to detect content which is on what format. Don't underestimate Google, opensourced TensorFlow is given by Google.

 

[world0]

Patterns detection is the way they detect AI the detectors. But they aren't good.

Regarding the watermarking thing, sound bad, but they will be workarounds around it.

 

[_Gray]

I think the deeper and more important question is not whether they can, because they could do it if they wanted to.

In my opinion, the question is do they care?

And my answer is no, I don't think they care. In fact, Google historically has always been a huge fan of automation. As long as the content is great, I don't see any logical reason to penalize it.

Think about it from their side. Google wants people to come back to their search engine so that they can generate money via ads. Their top-level strategy is to rank on the first page websites that provide epic UX, so that people get answers to their questions and continue using Google. The existence of AI content is being treated on the forum as "we are doing it the easy way, so there must be punishment coming our way". It's fundamentally wrong. Nobody cares if it was easy or hard. Only the result matters.

On that note, I'll just say that generating awesome content with AI is not easy. A really good article still requires manual research to have a solid structure. It requires creativity in the way the content is laid out. It needs lots of manual input to make the content 'beefy'. AI just doesn't know how to do some of the stuff well enough (yet), which means that unless you provide it with a lot of feedback, many articles will be low quality. Try using it as a writing assistant that you guide through every paragraph with your own creative ideas. Then polish it at the end a bit to give it personality.

 

[Elvlin]

I think the deeper and more important question is not whether they can, because they could do it if they wanted to.

In my opinion, the question is do they care?

And my answer is no, I don't think they care. In fact, Google historically has always been a huge fan of automation. As long as the content is great, I don't see any logical reason to penalize it.

Think about it from their side. Google wants people to come back to their search engine so that they can generate money via ads. Their top-level strategy is to rank on the first page websites that provide epic UX, so that people get answers to their questions and continue using Google. The existence of AI content is being treated on the forum as "we are doing it the easy way, so there must be punishment coming our way". It's fundamentally wrong. Nobody cares if it was easy or hard. Only the result matters.

On that note, I'll just say that generating awesome content with AI is not easy. A really good article still requires manual research to have a solid structure. It requires creativity in the way the content is laid out. It needs lots of manual input to make the content 'beefy'. AI just doesn't know how to do some of the stuff well enough (yet), which means that unless you provide it with a lot of feedback, many articles will be low quality. Try using it as a writing assistant that you guide through every paragraph with your own creative ideas. Then polish it at the end a bit to give it personality.

yup basicly most people here think ai content is prompt generate finish. no need to thinker, input data change setting etc. modiy etc.

 

[hellboy_1]

If big G think it's AI, THEN IT IS AI. Doesn't matter whether it's handwritten or not.

 

[sageshark]

After all, some human writers can produce absolute garbage content.

ha ha ...this is epic

 

[mr.lol]

i don't think they can detect all AI content. As the long as the content is good google will accept.

 

[MoonStarz]

Might be difficult to know in the case of really good AI because that's the idea right make it look human, but even if they know it, won't affect them at all, maybe the AI can create better content who knows.