I program AI professionally—this is not how AI works.

This response it gave is an arbitrary response based on what a human online would plausibly write in relation to the query, it has nothing to do with a personal analysis of its own system or training data.

You can't use GPT's own answers as some sort of evidence for information beyond its own black-box. These answers are completely fabricated based on statistical weights. With the right input you could get it to respond the exact opposite way or even confess that its creator was a leprechaun.

Now, the point of this post is still valid, GPT was trained on general internet data, and the internet by and large is filled with younger people, and younger people do typically lean more left (and also have immature/ignorant political views, by definition).

I program AI professionally—this is not how AI works.

This response it gave is an arbitrary response based on what a human online would plausibly write in relation to the query, it has nothing to do with a personal analysis of its own system or training data.

You can't use GPT's own answers as some sort of evidence for information beyond its own black-box. These answers are completely fabricated based on statistical weights. With the right input you could get it to respond the exact opposite way or even confess that its creator was a leprechaun.

Now, the point of this post is still valid, GPT was trained on general internet data, and the internet by and large is filled with younger people, and younger people do typically lean more left leaning (and also have immature/ignorant political views, by definition).