AI Predicts Earthquakes & OpenAI Fine-Tunes GPT-4 & more

🌋 AI Predicts Earthquakes with Unprecedented Accuracy

• Researchers at the University of Texas developed an AI that successfully predicted 70% of earthquakes during a trial in China, showcasing its potential for improving earthquake risk mitigation and preparedness.

• The AI was trained on a five-year database of seismic data and ranked first in an international competition, outperforming 600 other designs by accurately forecasting earthquakes within 200 miles of their actual locations.

• Future research aims to test the AI in regions with robust seismic networks, such as Texas and California, and to integrate it with physics-based models to enhance prediction accuracy globally.

🤖 Procreate CEO ‘Really F*cking’ Hates Generative AI

• James Cuda, CEO of Procreate, expressed strong opposition to generative AI, stating, “I really fucking hate generative AI,” and emphasized that the company will not incorporate such technology into its products, prioritizing human creativity instead.

• Procreate's official stance highlights concerns that generative AI is "ripping the humanity out of things" and is built on a foundation of theft, as it often utilizes copyrighted works from artists without permission for training purposes.

• Cuda's remarks have garnered support from many artists who appreciate Procreate's commitment to traditional artistic practices, contrasting with other companies like Adobe that are aggressively integrating AI features into their products.

🎨 Midjourney Unveils Game-Changing AI Image Editor on the Web

• Midjourney has launched a new unified AI image editor on its website, integrating various features such as inpainting and outpainting into a single interface, making it easier for users to edit AI-generated images.

• The updated editor introduces a virtual brush tool for more precise inpainting, replacing the previous selection tools, and is available to users who have created at least 10 images on the platform.

• This release comes amid ongoing legal challenges, including a class-action lawsuit from artists alleging copyright violations, but Midjourney continues to focus on enhancing user experience and community engagement.

🚀 OpenAI Rolls Out GPT-4 Fine-Tuning

• OpenAI has introduced fine-tuning capabilities for its GPT-4o model, allowing developers to create customized versions tailored for specific tasks, enhancing performance and reducing costs.

• Fine-tuning enables the adjustment of the pre-trained model with custom datasets, making it more efficient for specialized applications, such as serving as a professional tutor for programming courses.

• The service is available for all paid usage tiers, with costs set at $25 per million tokens for training and varying rates for input and output tokens, along with a temporary offer of free training tokens for developers.

🌍 The EU wants no corner of the digital sphere left untouched

• Since the implementation of the Digital Services Act (DSA) and the Digital Markets Act (DMA), the EU has intensified its regulatory actions against major tech companies, achieving significant victories such as compelling TikTok to remove an "addictive" feature from its app.

• Apple has emerged as a primary target of the EU's enforcement efforts, facing formal accusations of violating DMA rules and potential fines, while the EU continues to evaluate its compliance with new App Store changes.

• The EU's regulatory push is set to expand, with upcoming decisions regarding compliance for platforms like X (formerly Twitter) and possible inclusion of Telegram under stringent DSA rules, as the bloc aims to ensure comprehensive oversight of the digital landscape, including artificial intelligence.

LOOI Smartphone Desktop Robot

The LOOI Smartphone Desktop Robot makes your phone super smart with ChatGPT! This cool little robot can talk to you, take pictures, and help you get things done. It’s like having a fun, smart friend on your desk, making your workspace awesome.