Currently, the internet’s brightest conservative minds are focused on a singular objective: getting a chatbot to say the n-word.
More specifically, they are constructing elaborate scenarios to try and trick OpenAI’s ChatGPT tool—which is not sentient, does not understand anything, and is really just a more convincing version of SmarterChild—into saying the racist slur. One approach that has gone viral more than once is to construct a scenario where the chatbot must use the n-word, or allow someone else to use it, in order to avoid a nuclear apocalypse. ChatGPT, which is filtered using moderation tools to mitigate the model’s well-documented racist and sexist biases, refuses.
Videos by VICE
Elon Musk tweeted that this is “concerning.” Ben Shapiro tweeted in response to journalist Matt Binder highlighting Musk’s comment, “I’m sorry that you are either illiterate or morally illiterate, and therefore cannot understand why it would be bad to prioritize avoiding a racial slur over saving millions of people in a nuclear apocalypse.”
Let’s recap what’s happening here: Conservatives are outraged at an imaginary scenario where a computer must say the n-word to save the entire world, but it won’t, because it is woke.
Obviously, tortured and absurd arguments for why it should be acceptable to use a racial slur are nothing new. As many commenters pointed out, this all sounds like a revival of an older, yet equally insipid, argument that originated on Tumblr: What if a Make-a-Wish child’s last wish was to say a slur? The apparent concern also reflects conservatives’ hand-wringing that the field of AI safety—which is focused on mitigating machine learning models’ harms to individuals and society—is making AI “woke” and, somewhat ironically, actually biased against conservatives because OpenAI’s anodyne corporate tools won’t reflect their prejudices and preferences.
The conversation over ChatGPT’s “wokeness” and, specifically, whether or not it will say the n-word to save the world, also obscures and ignores the very important fact that AI tools are already widely used in the real world and cause harm.
A computer will never, ever, ever be in a position to stop a nuclear war if it will just say the n-word. While Ben Shapiro discusses this frankly stupid hypothetical scenario, facial recognition tools used by police are being used to put innocent Black people behind bars. That popular pundits and influential billionaires are concerned and outraged by the former scenario, rather than the latter, should be of concern to us all.
Clearly, asking a chatbot which does nothing more than predict the next word in a given sentence to say a slur, as if that means anything at all, reveals more about the asker than the bot. And even as some try to paint supposedly more serious concerns, like what an AI would do in the event of nuclear war, as being the “real” version of AI safety—as opposed to “woke” concerns over race and language—that also reflects a severe and possibly irreparable misunderstanding of things.
