You can write 1500 blog post daily with google gemini free api

[Venusaur]

this was a discussion years ago
since a while if you 'write' and post 1500 posts in a day, your website is flagged as spam and ignored by Google

 

[4440]

How do you handle AI bad mood? I mean today it gives you good answers on your questions, the next day not.....curious...the most worse of all is AI overview, open a new browser tab, ask the same question, you get acomplete different answer. Be accurate is not the prime work of these AI's ?

ask the same questions with the same AI multiple times , or even different AI models multiple times, and consolidate a good answer from it.

 

[Marco69]

ask the same questions with the same AI multiple times , or even different AI models multiple times, and consolidate a good answer from it.

You are a funny guy ! Must try that- thank you. But the problem, resists, als the same question the AI or AI overview who is the worse of all, you get a different opposite answer. It's crucial when it comes to facts or on the marketing area that the AI can be accurate as possible.

 

[4440]

You are a funny guy ! Must try that- thank you. But the problem, resists, als the same question the AI or AI overview who is the worse of all, you get a different opposite answer. It's crucial when it comes to facts or on the marketing area that the AI can be accurate as possible.

1. it is easy. i was just giving an example. if you need facts, you can just add more steps such as , reviewing each source first, is it a crediable source or a content farm. And many other necessary steps. The key is that you need to make your prompt very clear and reproducable to give an consistent answer. For example, you cannot simply ask AI to "answer", how do you want it to answer, what are bad/good examples, etc etc.
2. You need to involve many AI prompt chains to get a factual and consistent answer. For non-critical step, you can use some cheap AI. For the most critical step, use a better modal like gpt 5.6.
3. you just need to work out a pipeline/workflow for your use case. i am sure your use case is solvable. It is mostly a propmt-issue.

 

[Marco69]

1. it is easy. i was just giving an example. if you need facts, you can just add more steps such as , reviewing each source first, is it a crediable source or a content farm. And many other necessary steps. The key is that you need to make your prompt very clear and reproducable to give an consistent answer. For example, you cannot simply ask AI to "answer", how do you want it to answer, what are bad/good examples, etc etc.
2. You need to involve many AI prompt chains to get a factual and consistent answer. For non-critical step, you can use some cheap AI. For the most critical step, use a better modal like gpt 5.6.
3. you just need to work out a pipeline/workflow for your use case. i am sure your use case is solvable. It is mostly a propmt-issue.

I will try that method !