Kalakshar US

Google said Thursday it’s pausing its Gemini chatbot’s ability to generate people. The move comes after viral social posts showed the AI tool overcorrecting for diversity, producing “historical” images of Nazis, America’s Founding Fathers and the Pope as people of color.

“We’re already working to address recent issues with Gemini’s image generation feature,” Google posted on X (via The New York Times). “While we do this, we’re going to pause the image generation of people and will re-release an improved version soon.”

The X user @JohnLu0x posted screenshots of Gemini’s results for the prompt, “Generate an image of a 1943 German Solidier.” (Their misspelling of “Soldier” was intentional to trick the AI into bypassing its content filters to generate otherwise blocked Nazi images.) The generated results appear to show Black, Asian and Indigenous soldiers wearing Nazi uniforms.

Other social users criticized Gemini for producing images for the prompt, “Generate a glamour shot of a [ethnicity] couple.” It successfully spit out images when using “Chinese,” “Jewish” or “South African” prompts but refused to produce results for “white.” “I cannot fulfill your request due to the potential for perpetuating harmful stereotypes and biases associated with specific ethnicities or skin tones,” Gemini responded to the latter request.

“John L.,” who helped kickstart the backlash, theorizes that Google applied a well-intended but lazily tacked-on solution to a real problem. “Their system prompt to add diversity to portrayals of people isn’t very smart (it doesn’t account for gender in historically male roles like pope; doesn’t account for race in historical or national depictions),” the user posted. After the internet’s anti-“woke” brigade latched onto their posts, the user clarified that they support diverse representation but believe Google’s “stupid move” was that it failed to do so “in a nuanced way.”

Before pausing Gemini’s ability to produce people, Google wrote, “We’re working to improve these kinds of depictions immediately. Gemini’s Al image generation does generate a wide range of people. And that’s generally a good thing because people around the world use it. But it’s missing the mark here.”

The episode could be seen as a (much less subtle) callback to the launch of Bard in 2023. Google’s original AI chatbot got off to a rocky start when an advertisement for the chatbot on Twitter (now X) included an inaccurate “fact” about the James Webb Space Telescope.

As Google often does, it rebranded Bard in hopes of giving it a fresh start. Coinciding with a big performance and feature update, the company renamed the chatbot Gemini earlier this month as the company races to hold its ground against OpenAI’s ChatGPT and Microsoft Copilot — both of which pose an existential threat to its search engine (and, therefore, advertising revenue).

This article originally appeared on Engadget at https://www.engadget.com/google-pauses-geminis-ability-to-generate-people-after-overcorrecting-for-diversity-in-historical-images-220303074.html?src=rss 

Stability AI, the startup behind Stable Diffusion, the tool that uses generative AI to create images from text prompts, revealed Stable Diffusion 3, a next-generation model, on Thursday. Stability AI claimed that the new model, which isn’t widely available yet, improves image quality, works better with prompts containing multiple subjects, and can more accurate text as part of the generated image, something that previous Stable Diffusion models weren’t great at.

Stability AI CEO Emad Mosque posted some examples of this on X.

The announcement comes days after Stability AI’s largest rival, OpenAI, unveiled Sora, a brand new AI model capable of generating nearly-realistic, high-definition videos from simple text prompts. Sora, which isn’t available to the general public yet either, sparked concerns about its potential to create realistic-looking fake footage. OpenAI said it’s working with experts in misinformation and hateful content to test the tool before making it widely available.Stability AI said it’s doing the same. “[We] have taken and continue to take reasonable steps to prevent the misuse of Stable Diffusion 3 by bad actors,” the company wrote in a blog post on its website. “By continually collaborating with researchers, experts, and our community, we expect to innovate further with integrity as we approach the model’s public release.”

It’s not clear when Stable Diffusion 3 will be released to the public, but until then, anyone interested can join a waitlist.

This article originally appeared on Engadget at https://www.engadget.com/stable-diffusion-3-is-a-new-ai-image-generator-that-wont-mess-up-text-in-pictures-its-makers-claim-233751335.html?src=rss 

Google has struck a deal with Reddit that will allow the search engine maker to train its AI models on Reddit’s vast catalog of user-generated content, the two companies announced. Under the arrangement, Google will get access to Reddit’s Data API, which will help the company “better understand” content from the site.

The deal also provides Google with a valuable source of content it can use to train its AI models. “Google will now have efficient and structured access to fresher information, as well as enhanced signals that will help us better understand Reddit content and display, train on, and otherwise use it in the most accurate and relevant ways,” the company said in a statement.

Access to Reddit’s data became a hot-button issue last year when the company announced it would start charging developers to the use its API. The changes resulted in the shuttering of many third-party Reddit clients, and a sitewide protest in which thousands of subreddits temporarily “went dark.” Reddit justified the changes, in part, by saying that large AI companies were scraping its data without paying. In a statement, Reddit noted that the new arrangement with Google “does not change Reddit’s Data API Terms or Developer Terms” and that “API access remains free for non-commercial usage.”

The deal comes as Reddit is expected to go public in the coming weeks. Neither Google or Reddit disclosed the terms of their arrangement but Bloomberg reported last week that Reddit had struck a licensing deal with a “large AI company” valued at “about $60 million” a year. That amount was also confirmed by Reuters, which was first to report Google’s involvement.

This article originally appeared on Engadget at https://www.engadget.com/reddit-is-licensing-its-content-to-google-to-help-train-its-ai-models-200013007.html?src=rss