Nvidia AI chip sales stall in China due to US export controls

Key Takeaways

Nvidia's AI chip sales stall in China

Nvidia's AI chip sales in China have stalled due to US controls on exports of advanced technology. Local chipmakers like Huawei have gained ground, offering domestically designed chips. Nvidia's CEO, Jensen Huang, has acknowledged the US has lost its edge in China's advanced AI chips market. The US has imposed controls on exports of advanced technology, initially stalling sales of Nvidia's H200 AI chips in China.

Nvidia faces competition from Huawei in China

Nvidia's AI chip sales in China are facing significant challenges due to competition from local chipmakers like Huawei. The US controls on exports of advanced technology have limited Nvidia's ability to sell its high-end graphics processing units (GPUs) in China. Huawei's HiSilicon division has been at the forefront of the development of AI chips in China, and its products have been widely adopted by local companies.

Teaching AI to learn from mistakes

Researchers are exploring ways to teach AI to learn from its mistakes. The goal is to create AI that can learn from failures and improve over time. A new framework called On-Policy Reinforcement Learning from Failure (On-F) allows AI to learn from failures and receive constant feedback. This approach could be used in a range of applications, from robotics to autonomous vehicles.

Journalist documents AI experiment and parenting lessons

Journalist Joanna Stern spent a year integrating AI into her home and work activities. She tested various AI-powered devices, including chatbots, humanoid robots, and self-driving cars. Stern found that AI can be useful for administrative tasks, but it also reinforced the importance of human relationships. She documented her experiment in a book and highlighted the need for better AI design and user experience.

Canada pushes for AI safety and equity at UN

Canada is pushing for safe and equitable adoption of artificial intelligence at the United Nations. The country's ambassador to the UN, David Lametti, says AI governance is a significant priority for Canada's team. Lametti has worked on AI governance in the past and is committed to ensuring that AI is developed and used in a way that promotes safety and equity.

Rental car AI scanners flag alleged damage

Rental car companies are using AI-powered systems to inspect vehicles and flag alleged damage. However, the technology is raising concerns about accuracy and reliability. Some customers have reported being charged hundreds of dollars for damage they say they never caused. Experts warn that AI systems can make mistakes, especially if they are not properly trained or calibrated.

Q&A: What it takes to lead in AI

To lead in AI, agencies need to identify objectives and work backward from them. They should define priority use cases and establish governance early. Experimentation should be done after understanding objectives and ground rules. Leaders should think about AI as an operating model rather than just a technology.

AI can't solve brain problems without good data

Artificial intelligence can't solve brain problems without good data. Neuroscience has been struggling to integrate diverse data for decades. Standardized data and protocols are necessary for AI to work effectively. The brain's first foundation models are starting to appear in areas where data integration has matured.

DarkMoon: Open-source AI pentesting platform

DarkMoon is an open-source AI-powered pentesting platform that can perform security assessments end-to-end. It uses a large language model to plan and execute attacks, and can deliver an evidence-backed report at the end. The platform separates the model that thinks from the tools that act, and has a control layer that exposes an allow-list of approved tools and workflows.

Africa's AI challenge: execution over models

Africa's AI challenge is not building more models, but executing them in real-world environments. The continent's AI systems operate in complex environments with fragmented data, uneven infrastructure, and inconsistent connectivity. To succeed, AI systems must be designed for resilience and reliability, not just accuracy.

AI and juror deliberations: a new threat to fair trials

The increasing use of AI-generated content in the legal profession is raising concerns about fairness and justice. AI-generated errors in legal documents and filings can lead to delayed or denied justice, and even wrongful convictions. Jurors may be exposed to biased or inaccurate information during deliberations, which can also lead to unfair outcomes.

Sources

• Nvidia's AI chip sales in China stall, as local chipmakers like Huawei take the lead
• Nvidia's AI chip sales in China stall, as local chipmakers like Huawei take the lead
• Can robots “fail forward”? Teaching AI to learn from its mistakes
• Joanna Stern Documents Yearlong AI Experiment and Parenting Lessons
• UN ambassador Lametti: Canada pushing for safety, equity in artificial intelligence
• Rental car AI scanners flag alleged damage, leaving customers with surprise bills
• Q&A: What It Takes to Lead in AI
• AI can’t solve the brain without data that fit together
• DarkMoon: Open-source AI pentesting platform
• Africa doesn’t need more AI models; it needs AI execution systems
• A New Threat to a Fair Trial: Artificial Intelligence and Juror Deliberations