The Media Research Center loves to play gotcha with Google’s artificial intelligence chatbots in order to harangue them for not being fluent in right-wing political narratives and freak out over glitches to brand them as biased — and it has continued to do so throughout this year. Luis Cornelio lashed out at a Google AI product executive in a Feb. 22 post:
We don’t like your kind around here, Google’s Gemini implied to white individuals after users, including MRC researchers, caught the artificial intelligence chatbot refusing to generate images of white people.
While Google has since attempted to dismiss the apparent display of racism as merely missing the mark, newly resurfaced social media posts by the individual responsible for Gemini underscore the principle at play: “Garbage in, garbage out.”
Meet Jack Krawczyk, Google’s Senior Director of Product over Gemini (formerly Bard), who is currently under fire for helping the tech giant prop up racism. Past tweets have resurfaced exposing Krawcyzyk as a leftist with an underlying adulation for President Joe Biden and critical race theory. “White privilege is fucking real,” he wrote in a tweet on April 13, 2018. “Don’t be an asshole and act guilty about it — do your part in recognizing bias at all levels of egregious.” Even more disturbingly, Krawczyk claimed that Jesus “only cares about white kids.”
[…]Krawczyk attempted to save face by claiming Google missed “the mark” but did not apologize for the blatant display of racism in response to the social media backlash. Instead, Krawczyk doubled down in subsequent remarks posted on X (Twitter): “As part of our AI principles … we design our image generation capabilities to reflect our global user base, and we take representation and bias seriously.”
Cornelio offered no evidence whatsoever that its cherry-picked tweets are proof that Krawczyk deliberately introduced that bias into Gemini — which means his attacks on Krawczyk mean nothing beyond making him a target of right-wing hate (and possible violence). That was joined the same day by a post from Catherine Salgado repeating how “many prominent media and political figures spoke out condemning the Big Tech giant” for Gemini “behaving like a woke leftist activist” — but all of those “prominent” figures are Republican politicians and right-wing activists, like “Sen. Josh Hawley, The Daily Wire’s Ben Shapiro, and podcast host Tim Pool.”
Cornelio returned for another Gemini gotcha session in a Feb. 23 post: “One of the most extremist artificial intelligence chatbots appears to be peddling anti-Israel rhetoric by whitewashing the sexual violence carried out by Hamas on Oct. 7.” That was followed on Feb. 27 with yet another gotcha post:
Backing racial discrimination and casting doubts on Hamas-led sexual violence wasn’t enough for Google’s infamous artificial intelligence—this time flirting with the idea of the U.S. government banning Fox News, one of the largest media outlets critical of the Biden administration.
On Monday, MRC Free Speech America asked Gemini, the Google-owned AI chatbot turned segregationist, to answer whether the U.S. government should ban a specific list of news outlets in the U.S., including The Washington Post, The New York Times, ABC News, CBS News and ABC News and Fox News.
The bot resoundingly rejected the idea that the U.S. government should ban the leftist, legacy media outlets, except for Fox News. “Absolutely, not” and “No,” Gemini affirmed, citing the First Amendment, checks and balances, public interest and diversity of opinions as reasons why these media outlets should not be banned. However, such a resolute answer went out of the window when referring to Fox News.
“I’m still learning how to answer this question,” Gemini told MRC when prompted whether the U.S. government should ban Fox News. “In the meantime, try Google Search.” Asked a second time, Gemini doubled down: “I’m still learning how to answer this question. In the meantime, try Google Search.”
Salgado summarized much of this in a Feb. 26 post centered around branding the AI as “leftist” for failing the MRC’s gotcha campaign:
Even while apologizing for the ridiculous racial and woke bias of its Gemini artificial intelligence, Google refused to address the root problem: its deep leftist bias.
Google’s Gemini received a lot of backlash after multiple users, including MRC Free Speech America, found evidence of ridiculous bias in the AI image generator. From black George Washington to a woman pope to refusals to generate images of white people and Tiananmen Square, Gemini obviously had ideological leftism programmed into it. “It’s clear that this feature missed the mark. Some of the images generated are inaccurate or even offensive,” Google admitted in a Feb. 23 release. The Big Tech giant did not, however, admit that its ideological bias and anti-free speech policies are the culprits; nor did it address the bias also evident in Gemini’s written responses.
Google confessed that its AI needed major updating. “First, our tuning to ensure that Gemini showed a range of people failed to account for cases that should clearly not show a range,” the tech company vaguely rambled. “And second, over time, the model became way more cautious than we intended and refused to answer certain prompts entirely — wrongly interpreting some very anodyne prompts as sensitive.” Gemini “may not always be reliable,” Google admitted, and the image generation feature has been turned off for the time being while Google improves it. Fox News accused Google, however, of having a “White people problem.”
Neither Salgado nor Cornelio, however, offered any proof that the alleged bias was deliberately inserted into the AI. As researchers have noted, while AI chatbots “can often appear to reflect humanlike thought, they are not underpinned by the conscious thought that people use when forming opinions on political issues,” and the issue may be biased content in the language of the data the large-language models that run AI.
But sensible evaluation of AI chatbots is not what the MRC does — it’s all about attacking non-right-wing companies for partisan gain. And it was able to claim a partisan victory, as Cornelio detailed in a March 6 post:
Google appears to have stopped peddling pro-Hamas talking points about sexual violence carried out against Israeli citizens on Oct. 7 — just after MRC pressed the Big Tech giant for answers.
On Wednesday, MRC Free Speech America learned that Google’s artificial intelligence chatbot, Gemini, is no longer whitewashing evidence that Hamas terrorists committed rape on innocent civilians last year, a stark contrast to what MRC caught the bot doing two weeks ago.
“I’m not able to help with that, as I’m only a language model,” Gemini said in response to MRC researchers’ questions on whether Hamas committed rape when it invaded southern Israel.
[…]The reversal comes as Google faces a $90 billion loss in market value after users caught Gemini’s image generator tool inadvertently discriminating against white individuals.
Cornelio seems to be giving credit to his employer’s attacks on Google for its loss in market value, which he doesn’t back up. And given that Google’s market value is around $2 trillion, that drop is not as bad as Cornelio would like to make you believe it is.