Optimizing to Remove Political Bias from AI Models Removes Other Types of Bias
askrally.com
askrally.com
But also, if LLM’s are maximally truth seeking, wouldn’t you expect them to have different opinions than the population? I expect LLM’s could be more confident about climate change, given the amount of data it has access to.
From a linked paper:
“This isn't just an academic concern. As businesses increasingly turn to AI for market research, customer insights, and decision-making, these inherent biases could significantly skew results.”
What the fuck is AI for? It seems pretty useless if it’s just going to regurgitate the opinion of the median voter.
...and now you have realized why so many of us do not buy into the hype. Because that is all it does - regurgitate. Hey, there are valid use cases where that functionality is awesome. But LLMs have functionality limits. We're still in the time frame where people are figuring out what those limits are.