Automatic content creation with Bertha.ai

AI is getting bigger day by day. Autoblogging is the future. I think other AI content creators will also release their plugins very soon. Nowadays, it's difficult to understand how Google identifies quality content that's not AI generated.
 
Thanks for sharing.

>The Fastest Way To Create Content For Your WordPress Website

What does WordPress has to do with it? :)
Edit: Ah. It seems to be a plugin.

20,000 words for $30.

1,000 Words Per Month For FREE: I will sign up and let you know how it goes.
 
I never hear of Bertha.ai but I hear about other tools like these and some say can make the content like human or even that will beat 30$ writer or so....
 
Nowadays, it's difficult to understand how Google identifies quality content that's not AI generated.
GPT models are predicting the next token with a certain probability. If Google has access to the data-set used for training a specific model and if it has access to the AI model itself, it can calculate that probability and identify the content that has been generated by that model.

BUT if the AI model has been trained/ fine-tuned on a private data-set, there is no way for Google to differentiate the generated content from the human written content.

My personal opinion is that Google will use their AI to improve other aspects that will help them establish if the content is good (eg: did the content deliver what the reader was expecting? could this content be better?)
 
Last edited:
GPT models are predicting the next token with a certain probability. If Google has access to the data-set used for training a specific model and if it has access to the AI model itself, it can calculate that probability and identify the content that has been generated by that model.

BUT if the AI model has been trained/ fine-tuned on a private data-set, there is no way for Google to differentiate the generated content from the human written content.
This of course is true. Most of this AI NLP services providers spend huge amount of money in research. So they're doing a pretty good job at the of the day.
 
GPT models are predicting the next token with a certain probability. If Google has access to the data-set used for training a specific model and if it has access to the AI model itself, it can calculate that probability and identify the content that has been generated by that model.

BUT if the AI model has been trained/ fine-tuned on a private data-set, there is no way for Google to differentiate the generated content from the human written content.

My personal opinion is that Google will use their AI to improve other aspects that will help them establish if the content is good (eg: did the content deliver what the reader was expecting? could this content be better?)
hi Cognitive, how's your AI doing? i was reading one of the comments in Jarvis FB. they are working on their own model/data set, slowly separating from GPT 3.
 
GPT models are predicting the next token with a certain probability. If Google has access to the data-set used for training a specific model and if it has access to the AI model itself, it can calculate that probability and identify the content that has been generated by that model.

BUT if the AI model has been trained/ fine-tuned on a private data-set, there is no way for Google to differentiate the generated content from the human written content.

They use random sampling of the top X tokens (TopK & TopP) so calculating the probability of GPT3 generation though that method is close to impossible.

However, the random sampling methods used leads to statistical attacks which can help with detection. There are also a few other methods that lead to detection (writing tone variation)

This of course is true. Most of this AI NLP services providers spend huge amount of money in research. So they're doing a pretty good job at the of the day.

Most of them (all?) don't do any research at all, they just resell OpanAi's GPT3 model (with one-shot training or finetuned models) .
 
I'm my perspective. Jarvis outshines them all
They use random sampling of the top X tokens (TopK & TopP) so calculating the probability of GPT3 generation though that method is close to impossible.

However, the random sampling methods used leads to statistical attacks which can help with detection. There are also a few other methods that lead to detection (writing tone variation)



Most of them (all?) don't do any research at all, they just resell OpanAi's GPT3 model (with one-shot training or finetuned models) .
 
The problem with these AI tools is that the longer the content is the poorer the content becomes.
 
GPT models are predicting the next token with a certain probability. If Google has access to the data-set used for training a specific model and if it has access to the AI model itself, it can calculate that probability and identify the content that has been generated by that model.

BUT if the AI model has been trained/ fine-tuned on a private data-set, there is no way for Google to differentiate the generated content from the human written content.

My personal opinion is that Google will use their AI to improve other aspects that will help them establish if the content is good (eg: did the content deliver what the reader was expecting? could this content be better?)
Yep, bounce rate and dwell time will be more important in the future imo.
 
GPT models are predicting the next token with a certain probability. If Google has access to the data-set used for training a specific model and if it has access to the AI model itself, it can calculate that probability and identify the content that has been generated by that model.

BUT if the AI model has been trained/ fine-tuned on a private data-set, there is no way for Google to differentiate the generated content from the human written content.

My personal opinion is that Google will use their AI to improve other aspects that will help them establish if the content is good (eg: did the content deliver what the reader was expecting? could this content be better?)

+1

Google is not going to allow AI to ruin one of their key ranking factors.
 
Yep, bounce rate and dwell time will be more important in the future imo.

+1

Google is not going to allow AI to ruin one of their key ranking factors.

If good quality tools become available to the masses, then content quality will be more important than ever. What differentiates your content from others? Personal experience? Data? Data collected by you / your team?
 
i was reading one of the comments in Jarvis FB. they are working on their own model/data set, slowly separating from GPT 3.

I don't think Conversion.ai/ Jarvis have the money to create their own AI model :). Only the training costs for GPT-3 have been estimated to $12 million. But I'm sure they will need to fine-tune GPT-3 (and/ or any other similar model they might be using). If they've not done this 'till now, they're kind of late to the party. Obviously, fine-tuning is much much cheaper and it does wonders in some cases.

PS: check my sig :D
 
If good quality tools become available to the masses, then content quality will be more important than ever. What differentiates your content from others? Personal experience? Data? Data collected by you / your team?

Yeah, where it counts we have the best content in the game.
Our pricing both in outsourcing and doing it in-house is around $10 to $12 per 100 words.

User experience with Google is worth it.

I agree.
 
Just saw birtha landing page. It's not automatic.
It requires hand holding, pulling string.
It that what I need to do I will stick with Jarvis/conversation ai.
 
Bertha.ai is a WordPress plugin which brings A.I. writing help into your website content creation process.

Anyone tried it?

https://bertha.ai/
I wonder when we will get AI generated images/videos of AI influencers.
There is already robot influencers/3d model influencers on IG. I think there was one in Korea that nobody even realized was not human.
 
Back
Top