Connect with us
MARE BALTICUM Gaming & TECH Summit 2024

Uncategorized

Simulating Neurodegeneration and Aging in Artificial Intelligence Systems

Published

on

simulating-neurodegeneration-and-aging-in-artificial-intelligence-systems

 

In recent years, the advancement of artificial intelligence (AI) has enabled developers to create systems capable of emulating various human abilities, such as image recognition and question answering. However, unlike the human brain, which can deteriorate with age, AI systems typically maintain or even enhance their performance over time.
Researchers at the University of California, Irvine, have recently embarked on a study to mimic aging and biological neurodegeneration in AI agents. Their paper, which was pre-published on arXiv, has the potential to guide the development of innovative AI systems that utilize ‘artificial neurodegeneration’ to perform specific tasks.
Yu-Dai Tsai, co-author of the paper, shared that the inspiration for this study arose from discussions with Dr. Baldi and Dr. Pishgar, covering topics in neurodegeneration, learning, and AI safety. Additionally, Tsai’s personal experience with his father’s cognitive decline following brain trauma sparked new perspectives on the subject’s relevance to computer science, particularly in deep learning.
The researchers aimed to induce cognitive decline in AI agents to gain insights into complex systems, potentially improving their interpretability and security. They utilized IQ tests conducted by large language models (LLMs), specifically the LLaMA 2 model, to introduce the concept of ‘neural erosion.’ This deliberate erosion involved removing synapses or neurons or adding Gaussian noise during or after training, leading to a controlled decline in the LLMs’ performance.
Their findings revealed a distinct pattern of decline in AI systems, mirroring neurodegeneration observed in humans. As artificial synapses and neurons were removed, the AI systems first lost abstract thinking abilities, followed by mathematical degradation, and ultimately, linguistic skills. This pattern aligned with neurodegeneration patterns observed in humans.
Tsai emphasized that this study marks the beginning of a series, with plans to develop specific tests for AI systems and extend the emulation to other neural diseases and neurodiversity. Furthermore, they aim to apply their methods to enhance AI security and interpretability. While collaborations with neuroscientists are welcomed, their primary focus remains on exploring new frontiers in AI studies, rather than replicating human brain diseases.
Source: techxplore.com

The post Simulating Neurodegeneration and Aging in Artificial Intelligence Systems appeared first on HIPTHER Alerts.

Continue Reading
Advertisement
Stake.com

Uncategorized

EU’s new AI rules: Industry opposed to revealing guarded trade secrets

Published

on

eu’s-new-ai-rules:-industry-opposed-to-revealing-guarded-trade-secrets

 
New regulations in the European Union (EU) are set to compel companies to increase transparency regarding the data they use to train artificial intelligence (AI) systems, potentially unveiling closely guarded industry practices, reports the Times of India.
Since OpenAI, supported by Microsoft, introduced ChatGPT to the public 18 months ago, there has been a surge in public interest and investment in generative AI. This technology enables rapid generation of text, images, and audio content.
As the AI industry expands, concerns have emerged regarding how companies source data for training their models, particularly whether using content from popular books and movies without creators’ consent constitutes a breach of copyright.
The EU’s new AI Act, phased in over the next two years, mandates stricter regulations while allowing time for businesses to adjust to new requirements. Nevertheless, the practical implementation of these rules remains uncertain, notes the report.
Mandating “detailed summaries”
A contentious provision of the AI Act requires organizations deploying general-purpose AI models like ChatGPT to provide “detailed summaries” of the training data. The newly established AI Office plans to release a template for these summaries by early 2025 after consulting stakeholders. However, AI companies oppose disclosing their training data, arguing it as a trade secret that could unfairly benefit competitors if made public, the report reveals.
In the past year, major tech firms including Google, OpenAI, and Stability AI have faced lawsuits alleging unauthorized use of content for AI training. Despite US President Joe Biden’s executive orders addressing AI security risks, legal challenges regarding copyright remain largely untested, the report adds.
Backlash against OpenAI
Amid heightened scrutiny, tech companies have struck content-licensing deals with media outlets and websites. OpenAI, for instance, has partnered with the Financial Times and The Atlantic, while Google has collaborated with NewsCorp and Reddit.
Despite these efforts, OpenAI drew criticism in March when Chief Technology Officer Mira Murati declined to confirm whether YouTube videos were used to train its video-generating tool, Sora, citing potential violations of company terms and conditions.
Source: business-standard.com
The post EU’s new AI rules: Industry opposed to revealing guarded trade secrets appeared first on HIPTHER Alerts.

Continue Reading

Uncategorized

Picsart teams up with Getty to take on Adobe’s ‘commercially-safe’ AI

Published

on

picsart-teams-up-with-getty-to-take-on-adobe’s-‘commercially-safe’-ai

 
Picsart and Getty Images are joining forces to develop an AI image generator exclusively trained on Getty’s licensed stock content.
According to Picsart, their AI lab is constructing a bespoke model from the ground up to power this tool. It aims to provide the platform’s paying subscribers with the ability to generate images that come with full commercial rights. This initiative seeks to address persistent concerns about potential copyright violations associated with AI-generated content. The Picsart / Getty Images generator is slated for launch later this year and will be accessible through Picsart’s API services.
This collaboration bears similarities to Adobe’s Firefly AI model, initially introduced as a prompt-based image generation tool within Photoshop last year. Adobe has since expanded its integration across various Creative Cloud applications. Adobe’s model also emphasizes commercial safety by training on stock images from Adobe’s own library, along with openly licensed or out-of-copyright content. However, questions remain about the integrity of the training data and user trust in Adobe’s approach.
Getty Images has previously ventured into commercially-focused AI products through partnerships with Bria AI and Runway, and by teaming up with Nvidia to introduce “Generative AI by Getty Images,” leveraging its extensive catalog of licensed images. Adobe’s widespread integration of the Firefly model into popular applications like Photoshop, Illustrator, Lightroom, and Express may pose a challenge for Picsart’s new offering in terms of attracting creatives away from Adobe’s established ecosystem.
Source: theverge.com
The post Picsart teams up with Getty to take on Adobe’s ‘commercially-safe’ AI appeared first on HIPTHER Alerts.

Continue Reading

Uncategorized

Microsoft to delay release of Recall AI feature on security concerns

Published

on

microsoft-to-delay-release-of-recall-ai-feature-on-security-concerns

 

On June 13, Microsoft announced that it will postpone the rollout of its AI-powered feature “Recall” with new computers next week due to privacy concerns. Instead, the tech giant plans to offer Recall for preview to a smaller group later, following feedback and additional testing.
Recall is designed to track various activities from web browsing to voice chats, compiling a searchable history stored on the user’s computer. This allows users to easily retrieve past actions, even months later.
Originally slated for broad availability on June 18 for Copilot+ PC users, Recall will now undergo a preview phase exclusively within Microsoft’s Windows Insider Program (WIP) in the coming weeks. This decision, as stated in a blog post by the Redmond, Washington-based company, underscores their commitment to ensuring a trusted, secure, and reliable experience for all customers.
Copilot+ PCs, introduced in May, feature advanced AI capabilities aimed at enhancing user interactions and productivity. The WIP, a platform for software testing, enables enthusiasts to preview upcoming Windows operating system features.
Microsoft intends to incorporate feedback from the WIP community before extending the Recall preview to all Copilot+ PC users in the near future.
Following the feature’s announcement, concerns over privacy were swiftly voiced on social media, with some users fearing potential surveillance implications. Elon Musk, prominent technologist and billionaire, likened Recall to a scenario from the dystopian series “Black Mirror,” highlighting societal apprehensions about the impact of advanced technologies.
Source: reuters.com

The post Microsoft to delay release of Recall AI feature on security concerns appeared first on HIPTHER Alerts.

Continue Reading

Trending