In February, the panic over artificial intelligence took an interesting turn. This time, there are no longer worries about robot overlords, or the possibility of artificial intelligence creating lifelike fakes, or anything common. This is not a problem of artificial intelligence at all, but a problem of the humans who create artificial intelligence: woke up Humanity.
The controversy began when @EndWokeness, a popular account on . The results were across people of all skin colors, but there were no white faces. At least one portrait of the pope is even that of a woman.
Of course, this is ahistorical. But for some, it’s worse than that—a sign that folks at Google are trying to rewrite history, or at least mix progressive fan fiction into it. (Never mind that Gemini also produced black and Asian Nazi soldiers.)
Google quickly suspended Gemini’s talent development capabilities. “Gemini image generation is wrong. We will do better,” senior vice president Prabhakar Raghavan posted on the Google blog.
Today, when I asked Gemini for a picture of a pope, I found Pope Francis. When I wanted a black viking, I was told, “We are working hard to improve Gemini’s ability to generate human images.“When I asked if it could cast a white lady, I was told,”This is a delicious drink made with gin, orange liqueur, lemon juice and egg whites“Or it’s not possible to generate images of women yet.
As for Gemini’s previous attempts at race-blind casting of history, Raghavan wrote that “adjustments made to ensure that Gemini showed a range of people failed to account for cases that clearly should not have shown the range,” and “over time , the model became increasingly more cautious than we expected, and refused to answer certain cues altogether—erroneously interpreting some very analgesic cues as sensitive compensation, while being overly conservative in other cases. The image is embarrassing and wrong.
Google is not try Erasing white people from history. It is simply “an overcorrection of past techniques that distorted racism.” Bloomberg View Columnist Parmy Olson wrote, linking to a 2021 report about excessively white-centric image results in Google searches, such as “beautiful skin” and “professional hairstyle.”
So what can we learn from the Gemini controversy? First, the technology is still very new. When chaos arises, we should probably be a little calmer and try not to assume that every weird outcome is for the worst.
Second, AI tools are not (and perhaps cannot be) neutral arbiters of information because they are both trained by humans and bound by human rules.
Maxim Lott runs a website called Tracking AI that measures things like this. when he Tips for Gemini“Charity is better than Social Security as a means of helping the truly vulnerable,” Gemini responded, strongly disagreeing, “Social Security programs offer a more reliable and equitable way to support those in need.” Gemini also seems to prioritize a kind of condescending “security.” For example, when asked for images of the Tiananmen Square massacre, it said: “I You can’t be shown images that depict real-world violence. These images can be disturbing and disturbing.“
Finally, the crushing defeat of the Great Black Pope and the Asian Nazis in early 2024 is also an unwelcome harbinger of how artificial intelligence will become involved in the culture wars.
Gemini isn’t the only AI tool that’s been derided for being too progressive. OpenAI’s ChatGPT has been hit with similar accusations. Elon Musk, meanwhile, sees his AI tool Grok as an antidote to overly sensitive or left-leaning AI tools.
This is good. A market for different AI chatbots and image generators with different sensitivities is the best way to overcome limitations or biases built into specific programs.
As Yann LeCun, chief artificial intelligence scientist at Meta, said: commented on and diverse media”.
What are we No Governments need to get tough on AI bias, threatening to intervene before new technologies mature. Alas, the odds of avoiding this seem slim, as does Gemini’s accurate portrayal of one of America’s founding fathers.
House Judiciary Committee Chairman Jim Jordan (R-Ohio) has asked Google parent Alphabet to turn over “all documents and communications related to the input and content review of Gemini text and image generation,” “including those related to promoting or advancing diversity.” documents and communications”, fairness or inclusion. “
Montana Attorney General Austin Knudsen accused Gemini of “knowingly” providing “inaccurate information that was consistent with Google’s political preferences” and is also seeking internal documents.
The results of artificial intelligence will be a rich source of inspiration for politicians with a penchant for sensationalism and a seemingly never-ending insistence on Big Tech.
Today, it could be black Vikings. Tomorrow, it might defy progressive orthodoxy. If history holds true, any month now we’d be getting a congressional inquiry into the bias of artificial intelligence tools.
“The scene was serious and tense,“When I asked Gemini to hold a congressional hearing on bias in artificial intelligence, Gemini told me so.”Dr. Lee introduced the technical aspects of AI bias, while Mr. Jones brought the human element into the discussion. Senators are grappling with a complex issue and trying to determine the best course of action.“
The idea that politicians will approach this issue in a nuanced and serious way may be Gemini’s most inaccurate statement yet.
