Re:Conservatives Are Panicking About AI Bias, Think ChatGPT ...
by oceanplexian on Wednesday January 18, 2023 @10:53AM
(#34428251)
> Without that context the fact is useless and misrepresents reality.
I don't think that this is necessarily true.
The hypothetical fact in question- "At a certain point in history, Jewish people were overrepresented in finance" doesn't imply any bias. If this was a response provided by an AI it would be working as well as an encyclopedia or a history book, especially if the user asked for a concise answer. The computer didn't cause harm, the reader didn't cause harm, the language is neutral and to the point. The user has agency and can ask follow up questions.
The people assuming that the reader needs curated "context" are themselves biased. It's manipulative in the same way that voice assistants like Alexa try to manipulate you, for example, when you ask it to turn off the lights, it then volunteers "context" about how easy it is to buy products on Amazon.
I don't think that this is necessarily true.
The hypothetical fact in question- "At a certain point in history, Jewish people were overrepresented in finance" doesn't imply any bias. If this was a response provided by an AI it would be working as well as an encyclopedia or a history book, especially if the user asked for a concise answer. The computer didn't cause harm, the reader didn't cause harm, the language is neutral and to the point. The user has agency and can ask follow up questions.
The people assuming that the reader needs curated "context" are themselves biased. It's manipulative in the same way that voice assistants like Alexa try to manipulate you, for example, when you ask it to turn off the lights, it then volunteers "context" about how easy it is to buy products on Amazon.