😁Must-See Pixel 9 AI Leak

New Pixel 9 leak reveals AI feature ‘Add Me’ that ensures no one is left out of group shots

Key Points 

• This feature, shown in a now-removed ad, seems to stitch together two images with the same background, allowing someone to be added to a photo even if they were the one taking it.

• Other features include a new “Pixel Screenshots” tool that uses AI to help search through your screenshots. 

• Another leaked ad hints at an improved Magic Editor that could let you change a photo’s background based on a prompt.

📱News - A recent leak for the Pixel 9 has revealed a new AI feature called “Add Me.” This feature, shown in a now-removed ad, ensures no one gets left out of group photos.

The ad, which briefly appeared on YouTube, shows someone using a Pixel 9 to snap a photo of two friends in front of a van. After taking the picture, she swaps places with her friends to get her own shot in front of the van. The Pixel’s display shows ghostly images of her friends beside her, and when her friend takes the picture, she magically appears in the group photo as if she had been there the whole time.

Essentially, this feature seems to stitch together two images with the same background, allowing someone to be added to a photo even if they were the one taking it. It’s a neat feature, and it will be interesting to see it in action when Google officially launches the Pixel 9 on August 13th.

🤹🏻‍♂️What's more? Other features include Google’s Gemini assistant and a new “Pixel Screenshots” tool that uses AI to help search through your screenshots. Additionally, another leaked ad hints at an improved Magic Editor that could let you change a photo’s background based on a prompt.

😬In conclusion - These new features build on last year's updates from Google, like the Magic Eraser that removes people from photos and Best Take, which replaces blinking faces with better shots. However, they also blur the line between what’s real and what’s not, which could be problematic if misused.

Scientists warn AI systems may collapse into nonsense

Key Points 

• The surge in AI-created content due to excitement over systems like ChatGPT can create a feedback loop, where AI-generated text trains future systems, potentially leading to increasingly meaningless or nonsensical outputs.

• Researchers found that after a few cycles of generating and training on content, AI systems can produce nonsensical outputs; for instance, a system trained on medieval architecture text began generating repetitive lists of jackrabbits after just nine generations.

🙀News - Researchers have warned that as more content on the internet is generated by artificial intelligence, AI systems might "collapse" into producing nonsensical information.

🧐Let's understand why - Recent excitement around text-generating systems like OpenAI’s ChatGPT has led to a surge in AI-created content online. Many of these systems are trained using text from the internet, potentially creating a feedback loop where AI-generated content trains future AI systems. 

This cycle could result in these tools producing increasingly meaningless or nonsensical information. Researchers are warning that this phenomenon might contribute to the "dead internet theory," which posits that the web is becoming increasingly automated and self-referential.

🤖A case in point - Researchers found that after just a few cycles of generating and training on content, AI systems can start producing nonsensical outputs. For example, a system trained with medieval architecture text began producing repetitive lists of jackrabbits after only nine generations.

This issue is troubling as it not only makes the systems less useful but also reduces diversity in their outputs, potentially erasing smaller perspectives and varied viewpoints. 

👨🏻‍🔧What’s the fix? The problem could potentially be solved by watermarking AI outputs so they can be identified and excluded from training data. However, watermarks are easily removed, and AI companies have been hesitant to work together on this or other solutions.