NerdNewsAugust 12, 2025 |
News & Trends
The World Will Enter a 15-Year AI Dystopia in 2027, Former Google Exec Says
Former Google executive Mo Gawdat predicts a 15-year AI dystopia starting in 2027, citing AI's magnification of existing societal issues and 'human stupidity' as the primary drivers, rather than the technology itself. Gawdat believes humanity's values, such as freedom and human connection, will be disrupted, and that governments should regulate AI usage rather than the technology itself.
OpenAI's GPT-5 Debuts with Commoditizing Costs and Higher Scrutiny
OpenAI's GPT-5 features a router for longer thinking, new model sizes, and pricing aimed at production use. It also includes improvements in deception and benchmark performance, with a 400K token context and 128K maximum output tokens. The company's API surface has consolidated around the Responses API, with direct access to image generation and improved file search.
The GPT-5 Rollout Has Been a Big Mess
OpenAI's GPT-5 rollout has been marred by user complaints about broken workflows, lost AI models, and performance issues, prompting CEO Sam Altman to apologize and reverse key decisions.
Elon Musk Confirms Shutdown of Tesla Dojo
Elon Musk confirms Tesla has disbanded the team working on its Dojo AI training supercomputer, citing it as an 'evolutionary dead end'. The project's shutdown comes as Tesla shifts focus to its AI5 and AI6 chips, designed for inference and training. Musk announced the decision on social media, stating that resources will be focused on the new chips, which will be manufactured by TSMC and Samsung.
US demands cut of Nvidia sales in order to ship AI chips to China
The Trump administration has ordered Nvidia and AMD to pay a 15% commission on AI chip sales to China, potentially netting the US government $2 billion per year. The deal allows Nvidia and AMD to resume shipping AI chips to China after restrictions were put in place earlier this year. |
Options & Tutorials
Sandbox as a Service: Building an Automated AWS Sandbox Framework
This article proposes a scalable architecture for automated AWS sandbox environments, leveraging AWS Control Tower, AWS Organizations, and open-source tools like DCE and AWS Nuke. It enables secure, governed, and cost-efficient experimentation, while minimizing operational overhead and ensuring compliance with enterprise standards.
Figma Uses AI to Support Designers
Figma has integrated AI across its design platform, allowing for rapid prototyping and generation of production-ready code. The AI features build on existing design-to-code infrastructure, ensuring designers retain control over the final output. Figma's approach keeps AI as an assistant, speeding up work while leaving the craft and final call in human hands.
MSSQL Extension for VS Code 1.34.0
The Azure SQL team has released version 1.34.0 of the MSSQL extension for Visual Studio Code, improving Copilot Agent mode, adding colour-coded connections, and streamlining local container workflows. The update deepens the integration with GitHub Copilot Agent mode, allowing it to change databases, list schemas, and run queries directly from the chat assistant.
NVIDIA may give US government a cut of its profits to sell AI chips to China
NVIDIA and AMD have been granted export licenses to sell AI chips to China, but they must give the US government 15% of their profits. This move has sparked debate over national security concerns, with some experts warning it could endanger the US' economic and military edge in AI.
How to Build a Career That Thrives Alongside AI
The article discusses how to build a career that thrives alongside AI by becoming a T-shaped professional, with deep expertise in one area and broad capabilities across related domains. It highlights the need to develop skills that are scarce and complementary to AI, such as human-centric skills like teamwork, storytelling, and empathy. |
Launches & Tools
Claude AI Chatbot
Claude, an AI chatbot, now has the ability to reference past chats when asked, allowing users to pick up where they left off without re-explaining themselves. This feature is available to users on Max, Team, and Enterprise subscription plans and can be enabled or disabled via a settings toggle.
Nvidia gives its tiniest workstation GPUs a Blackwell boost
Nvidia has released two new workstation GPUs, the RTX Pro 4000 SFF and RTX Pro 2000, with a 70W TDP. The RTX Pro 4000 SFF has 8,960 CUDA cores and offers 1.7x faster ray tracing and 2.5x higher AI performance compared to its predecessor. The RTX Pro 2000 has 4,352 CUDA cores and promises a 1.6x improvement in 3D modeling and 1.4x higher performance in CAD.
Ford throws out Henry Ford’s assembly line to make low-cost EVs in America
Ford is abandoning its traditional assembly line to produce low-cost EVs in the US. The company will invest $2 billion to transform its Louisville Assembly Plant, introducing a new 'universal production system' that reduces production time and costs. The first EV to be produced will be a mid-sized pickup truck with a base price of $30,000, slated for launch in 2027.
Apple's New Siri
Apple is testing a new version of Siri that can take actions across various apps using voice commands, with plans to release a new version of App Intents, allowing developers to enable user access to parts of their app via different systems.
EDGX closes €2.3m funding to boost AI compute for satellites
EDGX, a Belgian space tech startup, has received €2.3m in funding to accelerate the commercialization of its edge AI computer for satellites, Sterna. Sterna is a high-performance data processing unit powered by NVIDIA technology, designed to provide computational performance and AI acceleration for satellites. This enables satellites to analyze and act on data in real-time, rather than waiting for ground processing. |
Quick Links
Pinterest Uncovers Rare Search Failure During Migration to Kubernetes
Pinterest engineers encountered a rare 'one-in-a-million' failure while migrating their search infrastructure to Kubernetes. The issue was caused by subtle inconsistencies in containerized search components and legacy infrastructure, triggered by a rare timing condition. The team used incremental isolation, custom logging, and traffic replay to identify the anomaly, highlighting the importance of robust observability and debugging in distributed systems.
The biggest challenges in software supply chains in 2025
The article discusses the biggest challenges in software supply chains in 2025, including open source, supply chains, AI, and more. It highlights the need for secure open source tooling and maintenance, and recommends four actions to meet these challenges: exploring open source technologies, bringing the software supply chain into the core of software delivery, automating OS updates, and identifying the impact of regulatory and compliance requirements.
AI Security
Experts discuss the role of AI in security, with some believing it currently favors defenders over attackers, while others predict offense will gain the upper hand soon. AI is being used for red teaming and defense, but its effectiveness is limited without human direction.
Ford's New Way of Making Cars
Ford is introducing a new way of making electric vehicles, called the Ford Universal EV Production System. This method involves building the car in three separate parts - front, middle, and rear - and then assembling them together. This approach is expected to be 40% faster and reduce the number of workstations and parts needed. The company is spending $2 billion to set up this new system at its Louisville assembly plant.
What Does Palantir Actually Do?
Palantir is a software company that provides data integration and analysis tools to government agencies and corporations. Its software, including Foundry and Gotham, helps customers connect and transform different kinds of data, making it easier to analyze and make decisions. However, the company's work with government agencies, such as Immigration and Customs Enforcement, has sparked controversy and protests. |
Share NerdNewsShare your affiliate link to get commission!
Thanks for reading,
If you dont want to receive future editions of NerdNews, |