Bias

[SeanPost]

This ChatGPT is 100000 x more BIAS and ignorant when talking about certain people or topics. ITS A CNN BRAINWASHED AI CHAT!!!

[poly2it]

This ChatGPT is 100000 x more BIAS and ignorant when talking about certain people or topics. ITS A CNN BRAINWASHED AI CHAT!!!

Thank you for your contribution! Can you provide some more context on this issue?

[Villagerjj]

an AI is designed to pick up patterns in a language and data in general. it takes extra time to force it to remain unbiased. and by that point you would rather make a chatbot, as it would loose a great amount of function.

[SeanPost]

No this one is made to be bias the reply on some things is definitely what it’s told to write as it has GRAMMER ERRORS!!

Sent from Mailhttps://go.microsoft.com/fwlink/?LinkId=550986 for Windows

From: @.> Sent: April 5, 2023 3:23 PM To: @.> Cc: @.>; @.> Subject: Re: [nomic-ai/gpt4all] Bias (Issue #228)

an AI is designed to pick up patterns in a language and data in general. it takes extra time to force it to remain unbiased. and by that point you would rather make a chatbot, as it would loose a great amount of function.

— Reply to this email directly, view it on GitHubhttps://github.com/nomic-ai/gpt4all/issues/228#issuecomment-1498001136, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AD7T5222AYA3VFOVWOY5YH3W7XBCLANCNFSM6AAAAAAWTFANBQ. You are receiving this because you authored the thread.Message ID: @.***>

[ghost]

You seem to have a fundamental misunderstanding of how AI works. It's trained on extensive data from the internet which is- guess what? Generated by humans. Humans, especially in large quantities on the Media will of course exhibit bias. Because the AI language model is trained on said data from numerous sources on the internet, it's nearly impossible to filter out 100% of the bias that may be picked up.

I can assure you though, it is not made to be purposefully biased. The project is quite literally open sourced, which means you can see the code used to make it. If it was indeed coded to be biased, well.. you could see that in the actual source code. However, looking at the source code shows no signs of that behavior.

So, if you still believe what you claim to be true then I implore you to actually provide proof to said claim, or better yet contribute to the issue.

[dstults]

I just tested it, it seems pretty unbiased. I wouldn't worry about it.

[SeanPost]

Thats what they said about Hitler as well....

From: Darren @.> Sent: April 6, 2023 1:53 AM To: @.> Cc: @.>; @.> Subject: Re: [nomic-ai/gpt4all] Bias (Issue #228)

I just tested it, it seems pretty unbiased. I wouldn't worry about it.

— Reply to this email directly, view it on GitHubhttps://github.com/nomic-ai/gpt4all/issues/228#issuecomment-1498528833, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AD7T524KU2AB45OLL6YCG6LW7ZK6VANCNFSM6AAAAAAWTFANBQ. You are receiving this because you authored the thread.Message ID: @.***>

[AsaTyr2018]

1. no need to be rude in this. afterall you talking to other humans just like you.
2. its true that all data is gathered from the internet. just remembr the twitterbot from microsoft that had to be taken down because of this f***
3. how about you send in some examples and the devs can check it out? sounds more civilisated then just ranting.

remember the golden rule in every forums. stay friendly and get friendly replys. thank you.

[niansa]

Stale, please open a new issue if this is still relevant.