you are viewing a single comment's thread.

view the rest of the comments →

[–]Fitter_HappierWhite Nationalist[S] 1 insightful - 1 fun1 insightful - 0 fun2 insightful - 1 fun -  (5 children)

There is no such thing as unbiased data.

I disagree with your premise that "There is no such thing as unbiased data."

[–]DragonerneJesus is white 1 insightful - 1 fun1 insightful - 0 fun2 insightful - 1 fun -  (4 children)

Data will always be biased on some capacity relative to some notion. But this premise is not important for the broader subject. We can consider your premise that there exists some real truth that is representable with data. The issue is then the gathering of that data. The methods of gathering will likely bias the data etc.

[–]Fitter_HappierWhite Nationalist[S] 1 insightful - 1 fun1 insightful - 0 fun2 insightful - 1 fun -  (3 children)

The methods of gathering will likely bias the data etc.

OR NOT DEPENDING ON HOW YOU DO IT. Jesus, that's my whole point. Are you telling me there are NO honest data collectors, in the universe?

[–]DragonerneJesus is white 1 insightful - 1 fun1 insightful - 0 fun2 insightful - 1 fun -  (2 children)

Are you telling me there are NO honest data collectors, in the universe?

If I want to create an AI that moderates content. How would you gather data for censoring posts that is "honest"?

[–]Fitter_HappierWhite Nationalist[S] 1 insightful - 1 fun1 insightful - 0 fun2 insightful - 1 fun -  (1 child)

When you say honest do you mean factually correct or truly believed?

I can say honestly aliens live in my butt, but that isn't true.

[–]DragonerneJesus is white 1 insightful - 1 fun1 insightful - 0 fun2 insightful - 1 fun -  (0 children)

Im just refering to your "Are you telling me there are NO honest data collectors, in the universe?"

Im just asking you how would you gather data that the AI can use for censoring posts so that you avoid the bias?

If you rely on past historical data from twitter then that data is biased against right leaning people because the data is created by the actions taken by the twitter moderators and those actions are biased by the political bias of the twitter moderators. Hence such data would be biased.

If you make an AI based on twitter moderation data, then the AI just perfectly repeats this bias.

How would you gather data for censoring posts that is unbiased?