[FREE TOOL]Cointikka: Tools for Bitcoin Holders & Bitcointalk Users

[joniboini]

I am checking for other reliable models to replace it with, but I have not found one better than it. Model 1 is perfect and does not have any issues that is why I am now giving options to users on whether they want to use Model 2.

Maybe you can add an option for the users to communicate with their local LLM as I mentioned before? For example I have an Ollama with a gemma model. I can use the website to scrap texts from the forum, then throw them to my LLM to check, and show the result on the website in real time. I know gemma isn't the best model for testing AI texts, but this is just an example. This means users don't have to download the model manually and have more control over which models they wanted to use.

[Cointikka]

Maybe you can add an option for the users to communicate with their local LLM as I mentioned before? For example I have an Ollama with a gemma model. I can use the website to scrap texts from the forum, then throw them to my LLM to check, and show the result on the website in real time. I know gemma isn't the best model for testing AI texts, but this is just an example. This means users don't have to download the model manually and have more control over which models they wanted to use.

I will look into your suggestion. If I am able to implement what you are suggesting, then I will give it as another option to the user. BTW.. if you are using Chrome, you can paste any text into the integrated Gemini AI to check whether an AI has written it. It will give you factual information, but it will not be able to give the percentage of text being written by the AI.

[joniboini]

if you are using Chrome, you can paste any text into the integrated Gemini AI to check whether an AI has written it. It will give you factual information, but it will not be able to give the percentage of text being written by the AI.

I'm not really a fan of Gemini AI or anything being solved into my face in general, but I do have an extension to communicate with my LLM directly inside my browser. It's definitely a quick way to check whether I'm consuming AI texts or not, but having a shareable result is better. I think your tool can be one of that, assuming you can nail down the accuracy to an acceptable result. As in removing model that always gives contradictory results or unreliable judgements.