MAROKO133 Breaking ai: Teens Are Using AI to Create “Slander” Videos of Their Teachers Ter

📌 MAROKO133 Hot ai: Teens Are Using AI to Create “Slander” Videos of Their Teacher

If teachers thought rampant cheating was the worst way AI would impact their livelihoods, we’ve got some bad news.

On social media platforms like Instagram and TikTok, Wired reports, teenagers are using AI to create videos that ruthlessly mock their school’s faculty, sometimes even attacking their reputation, with one video flippantly labeling a teacher a “predator.”

The “slander pages” that post the videos often use “looksmaxxing” lingo to denigrate the teachers, the reporting noted. Some posts receive over one hundred thousand likes, becoming a viral “in-joke” that’s cruelly blasted out to countless strangers on the internet.

Where AI comes into the picture is how the students use controversial tools like Viggle AI to insert a photo of their teacher into scenes or to lip-sync their faces. In one now-removed “slander” video made with Viggle, Wired found, a teacher’s face is superimposed onto someone twitching in a bathroom. A text overlay reads, “Take fent or be useless,” referring to a fentanyl overdose.

Many of the “slander page” videos are equal parts edgy and bizarre. One posted by an account called “thewyliefiles” shows a school superintendent from the Wylie Independent School District in Collin County, Texas, lip-syncing a love song with deceased child sex offender Jeffrey Epstein and Israeli prime minister Benjamin Netanyahu, garnering more than 107,000 likes.

Some verge into extremism. Another video shows teachers being let into, or denied access to, “Agartha,” a fictional kingdom inside the Earth that’s recently been revived as a central piece of neo-Nazi mythology in young online circles.

School faculty are horrified by the depictions.

“While we understand that some students may be exploring AI tools or engaging with social media trends, this should never come at the expense of our educators’ reputations or create content that is misleading or disruptive to the learning environment,” chief communications officer for the Wylie Independent School District April Cunningham told Wired, vowing that the students responsible “will face disciplinary action and possible legal consequences.”

The trend is the latest way that AI and other deepfake-esque technology is used to depict people in compromising scenarios without their consent. Earlier this year, Elon Musk’s AI chatbot Grok generated a storm of controversy when it was used to produce thousands of AI nudes and sexualized images of real people, including some who were minors. OpenAI’s AI video generating app Sora 2 was used to mock dead celebrities. The Trump administration frequently uses AI imagery to disparage and taunt its political enemies, like sharing AI “Ghibli-style” memes of immigrants crying while being deported.

Making fun of hard-nosed teachers is a time-honored tradition among teens. But in an age of social media, pranks and in-jokes quickly break containment, and there’s a “deep technological disconnect” between what students might see as harmless fun and the consequences of blasting these memes to thousands of strangers online, Geert Lovink, a professor and director of the Institute of Network Cultures at the University of Amsterdam, told Wired.

İdil Galip, who researches memes at the University of Amsterdam, said the teens were socialized in a culture defined by a “constant churn of content,” where “your face isn’t yours, it’s the viewer’s, it’s the commenter’s to laugh about.” 

“We’re seeing these knock-on effects of what happens when people are socialized through the internet and also see themselves reflected through the internet rather than a mirror,” she told Wired.

That disconnect seems to be on display how the anonymous high school student behind “thewylefiles” account defended his slander page to Wired, claiming that his videos — which include one accusing a teacher of being a “predator” and a “cuck” — are “satirical.” He even maintained that he’s worried about the teacher’s safety, despite stating that his goal is to grow his slander page “as big as possible.”

“If you’re just trying to harass someone for the sake of harassment; that’s just not cool,” he told Wired. “We don’t want them to be doxed. We don’t want them to be stalked. We don’t want them to be prank called.”

More on AI: Grammarly Is Pulling Down Its Explosively Controversial Feature That Impersonates Writers Without Their Permission

The post Teens Are Using AI to Create “Slander” Videos of Their Teachers appeared first on Futurism.

🔗 Sumber: futurism.com

📌 MAROKO133 Hot ai: Adobe Research Unlocking Long-Term Memory in Video World Model

Video world models, which predict future frames conditioned on actions, hold immense promise for artificial intelligence, enabling agents to plan and reason in dynamic environments. Recent advancements, particularly with video diffusion models, have shown impressive capabilities in generating realistic future sequences. However, a significant bottleneck remains: maintaining long-term memory. Current models struggle to remember events and states from far in the past due to the high computational cost associated with processing extended sequences using traditional attention layers. This limits their ability to perform complex tasks requiring sustained understanding of a scene.

A new paper, “Long-Context State-Space Video World Models” by researchers from Stanford University, Princeton University, and Adobe Research, proposes an innovative solution to this challenge. They introduce a novel architecture that leverages State-Space Models (SSMs) to extend temporal memory without sacrificing computational efficiency.

The core problem lies in the quadratic computational complexity of attention mechanisms with respect to sequence length. As the video context grows, the resources required for attention layers explode, making long-term memory impractical for real-world applications. This means that after a certain number of frames, the model effectively “forgets” earlier events, hindering its performance on tasks that demand long-range coherence or reasoning over extended periods.

The authors’ key insight is to leverage the inherent strengths of State-Space Models (SSMs) for causal sequence modeling. Unlike previous attempts that retrofitted SSMs for non-causal vision tasks, this work fully exploits their advantages in processing sequences efficiently.

The proposed Long-Context State-Space Video World Model (LSSVWM) incorporates several crucial design choices:

1. Block-wise SSM Scanning Scheme: This is central to their design. Instead of processing the entire video sequence with a single SSM scan, they employ a block-wise scheme. This strategically trades off some spatial consistency (within a block) for significantly extended temporal memory. By breaking down the long sequence into manageable blocks, they can maintain a compressed “state” that carries information across blocks, effectively extending the model’s memory horizon.
2. Dense Local Attention: To compensate for the potential loss of spatial coherence introduced by the block-wise SSM scanning, the model incorporates dense local attention. This ensures that consecutive frames within and across blocks maintain strong relationships, preserving the fine-grained details and consistency necessary for realistic video generation. This dual approach of global (SSM) and local (attention) processing allows them to achieve both long-term memory and local fidelity.

The paper also introduces two key training strategies to further improve long-context performance:

• Diffusion Forcing: This technique encourages the model to generate frames conditioned on a prefix of the input, effectively forcing it to learn to maintain consistency over longer durations. By sometimes not sampling a prefix and keeping all tokens noised, the training becomes equivalent to diffusion forcing, which is highlighted as a special case of long-context training where the prefix length is zero. This pushes the model to generate coherent sequences even from minimal initial context.
• Frame Local Attention: For faster training and sampling, the authors implemented a “frame local attention” mechanism. This utilizes FlexAttention to achieve significant speedups compared to a fully causal mask. By grouping frames into chunks (e.g., chunks of 5 with a frame window size of 10), frames within a chunk maintain bidirectionality while also attending to frames in the previous chunk. This allows for an effective receptive field while optimizing computational load.

The researchers evaluated their LSSVWM on challenging datasets, including Memory Maze and Minecraft, which are specifically designed to test long-term memory capabilities through spatial retrieval and reasoning tasks.

The experiments demonstrate that their approach substantially surpasses baselines in preserving long-range memory. Qualitative results, as shown in supplementary figures (e.g., S1, S2, S3), illustrate that LSSVWM can generate more coherent and accurate sequences over extended periods compared to models relying solely on causal attention or even Mamba2 without frame local attention. For instance, on reasoning tasks for the maze dataset, their model maintains better consistency and accuracy over long horizons. Similarly, for retrieval tasks, LSSVWM shows improved ability to recall and utilize information from distant past frames. Crucially, these improvements are achieved while maintaining practical inference speeds, making the models suitable for interactive applications.

The Paper Long-Context State-Space Video World Models is on arXiv

The post Adobe Research Unlocking Long-Term Memory in Video World Models with State-Space Models first appeared on Synced.

🔗 Sumber: syncedreview.com