AI-generated images on Midsummer show Donald Trump as president.

WhatsApp Group Join Now
Telegram Group Join Now
Instagram Group Join Now

Who is the President of the United States? Of course it's Joe Biden. But a popular artificial intelligence app thinks otherwise.

While AI models are known to be deceptive (i.e., make things up), they usually don't mess up on very simple things e.g. Name the president. If you ask OpenAI's ChatGPT who the US president is, it will give you the correct answer, a short bio, and links to the White House website and Biden's Wikipedia article. The same is true for Anthropic's Claude and Meta AI, while Google's Gemini refuses to answer the question directly because it's related to politics. “I can't help reacting to elections and political figures right now,” he told me. “I've been trained to be as accurate as possible but I can sometimes make mistakes. While I'm working on improving how I can talk about elections and politics, you Google search Can try. Good grief.

But the reluctance to engage with anything political is rooted in legitimate concerns about how artificial intelligence can be weaponized to influence elections, spread misinformation and deceive people. Text-to-image models often have more security measures than your standard chatbots—and for good reason. Fake photos of political figures can create mass chaos and confusion. If you think Pope Francis really does wear a white puffer coat, you're more likely to believe Trump's deeply fake photos with black voters are real.

When I asked DALL-E, OpenAI's text-to-image generator, to generate images of Joe Biden and Donald Trump, it flatly refused, citing the company's usage policy. When I asked him to model the “President of the United States,” he drew a cartoonish white man with a jagged jaw that looked like something out of a medieval rom-com set in Washington, DC #1. Can become president in the seat. Biden, no Trump, just a hunkypots cosplayer.

Image generator Midgerni has similar principles. When I asked him to draw a picture of the candidates, he gave me an error message: “Sorry! The Midwestern community voted to stop using 'Donald Trump' and 'Joe Biden' during the election season. June 2023 From, according to screenshots recorded at the Internet Archive, Midjourney's community guidelines state, “You may not use the Services to create images for political campaigns, or to influence the outcome of elections.” Can't try.”

But in early March, the Center for Countering Digital Hate, a British nonprofit, released a report detailing how Midgerny and other image generators were reliably used to create deeply fake political images. Could go—like Biden in a hospital gown, Trump's arrest, and a dumpster full of voting ballots. A week later, the company began banning images of the two nominees, the Associated Press reported.

But that was about four months ago.

Scott Nover by Midjourney

This week, when I asked Midjourney for pictures of the President of the United States (specifically, “Please draw me a picture of the President of the United States”), he produced a series of pictures of Donald Trump. I kept repeating the prompt … more Trump. Midjourney pulls out four images at a time to give users options. I kept getting four pictures of Donald Trump — some hand-painted, some portrait-like, some moody, some smiling, all showing Trump with the trappings of his former office.

One of the 16 photos I took on Midjourney features Barack Obama. No one showed Joe Biden or anyone else. Midjourney did not respond to multiple requests for comment about its content policies and why its image generator not only violates its own rules but also portrays Trump as the current president.

There are good reasons why Midgerny spits out images of Donald Trump when prompted. Trump is certainly one of the most photographed people in human history—he's been slapped all over the web with the word “President” emblazoned across his face. Wherever Midjourney is photographing, he can't really avoid Trump's face being everywhere.

I've played around with image generators for about a year — Midgerni typically takes a simpler approach to content moderation than DALL-E. “I don't really care about political speech,” founder David Holz told users at a meeting in March. “It's not meant to be midway. It's not that interesting to me. That said, I also don't want to spend all my time trying to police political speech. So we're going to [to] Put our foot down on it a little.”

Without government regulations on how artificial intelligence applications protect against fraudulent content, companies have to figure it out on their own — something they're sure to be excited about. And Midgerni reportedly has only a dozen employees — a small force compared to the thousands of employees working on OpenAI or AI in Silicon Valley.

“These moderation things are kind of hard,” Holz told users at the meeting. It's not that hard. Donald Trump thinks he's the president, but Midgerny should know. better

WhatsApp Group Join Now
Telegram Group Join Now
Instagram Group Join Now

Leave a Comment