What the hell is a “DEI” statement anyways? The site is a bit light on what this means, only a bunch of satirical and sarcastic posts. Almost feels like it was generated by LLM.
Is it something like this?
“I helped a group of first generation, low income students apply to college and obtain $50,000 in grants”
Note how saying you’ll “treat everyone the same” will get you penalized. How doing the things they want isn’t enough to demonstrate your commitment to their cause.
They want true believers, but will settle for people who are willing to let others control what they say and do.
During the red scare, there was a movement to try and force professors to sign loyalty oaths stating that they weren’t communist. Imagine that this went one step further: rather than signing a loyalty pledge, they required professors to compose their own loyalty statements. And that they would be judged on the ideological correctness of their statement. For instance, it wouldn’t be enough to pledge not to be communist; they would also had to pledge to be actively anti-communist. This didn’t happen during the red scare, but if you s/communist/racist, it is happening now.
what? an implementation of an ai? no not really. Those things are hard, take millions of dollars to train ... it doesn't really show, say Marx's M-C-M' equation, r = s/(c + v) or any of the other theories in his text or anything else in that comment. I don't see how say https://www.marxists.org/archive/marx/works/1894-c3 relates to a place "with no dissent".
That's just a bunch of trigger-words and scapegoats pasted together.
You seem very confused about how Google Gemini, or any ML-based system for that matter, is trained.
It is a natural consequence of these systems that they output the same biases that go into the input; e.g., the classic translation of "they are a doctor/nurse" in a language with gender-neutral pro/nouns to "he/she is a doctor/nurse" in English. Therefore, if one were to train any such system on the Founding Fathers or Nazi soldiers, it is only natural to expect that the system would output white/German/Eastern European men in each case. Yet Gemini doesn't; instead, Gemini introduces its own bias to erase the one in the input and instead always produce an "equal" cocktail of white/black/asian men/women (which, ironically, is still biased and racist because that only captures the races that woke employees at Google feel are important, but it takes actual intellectual honesty to see one's own biases), regardless of whether that results in historical inaccuracies or downright falsehoods. This is a very intentional bias and dishonesty, not something that can just be shrugged off as "it's hard and takes millions of dollars to train".
You're looking at buggy output, attributing political agency to it and speculating some fanciful story about a moustache twirling blue haired villain burning tiny american flags.
It was a software defect.
It's like claiming Stability AI wants everyone to have 6 or 7 fingers because that's how it draws hands or they're secretly occult theosophists because their mirrors don't generate reflections right so they believe mirrors are dimensional portals.
Is it something like this?
“I helped a group of first generation, low income students apply to college and obtain $50,000 in grants”
Not sure what’s wrong with this…