Prices of Nvidia’s latest GPUs soar in China due to US export ban • The Register

Overview of AI Nvidia’s 2nd Gen A100 GPU is currently subject to export controls and is not available in China, but could fetch up to $20,000 on the black market, double the normal price.

Last year, the Commerce Department blocked top U.S. hardware companies, including Nvidia and AMD, from selling cutting-edge chips to undermine the technological prowess of China and Russia. Under the new rules, Nvidia has been banned from exporting his A100 and H100 GPUs.

but, Reuters The report reveals that Chinese developers can still source small amounts of A100 by importing it from other Asian countries such as India, Taiwan and Singapore on the black market and some electronics retailers. . Reportedly, one GPU in Shenzhen could cost him $20,000 if you ask carefully at specialty stores.

Iwan Lau, co-founder of the Pantheon Institute in Hong Kong, said he plans to buy two to four A100 chips to develop AI models. He said he was in talks with two vendors who purchased the device outside the United States, setting the price at HK$150,000 (about $19,150) per card. “They told us bluntly that they had no guarantee or support,” he said.

The sale of such chips is not illegal in China, but is frowned upon. US companies aren’t allowed to export their models directly, making them hard to find in large quantities, making it even more difficult for Chinese developers to train large and powerful models.

“If we receive information that a customer is in breach of their contract with us and exporting restricted products in violation of the law, Nvidia will immediately take appropriate action,” Nvidia said in a statement.

US Government Misses Opportunity to Address AI-Generated Political Ads

The Federal Election Commission is debating whether political ads created using generative AI tools should be banned. In its petition, the FEC was asked to clarify whether potentially misleading AI-generated ads could be considered “misrepresenting the authority of a campaign.” Half of the six-member committee voted in favor of meeting the petitioners’ demands, while the rest opposed any action.

“The Commission discussed an available notice requesting comment on whether the Commission should initiate full rulemaking based on proposals in rulemaking petitions from citizens, but by the four affirmative votes required. We did not authorize the publication of the notice,” the regulator admitted. This week’s somewhat complicated statement.

Concerns that AI-generated content can spread disinformation and undermine democracy in order to manipulate people’s views of political candidates are real. The Republican Party recently released a fake anti-Joe Biden ad made with AI, and another made for Ron DeSantis’ campaign targeted his main rival, Donald Trump. was.

Should these be regulated or outright banned? It’s not clear how the authorities plan to tackle the issue, nor are owners of social media platforms like Meta convinced. I can’t seem to have

Senator Chuck Schumer Announces AI Regulatory Framework

Senate Majority Leader Chuck Schumer this week announced an AI policy called SAFE Innovation in the Age of AI.

Schumer said at an event hosted by the think-tank Institute for Strategic and International Studies that the regulatory framework is aimed at mitigating AI safety risks while supporting innovation in the United States.

“Our framework must not lose sight of what should be our north star: innovation,” he said in his speech. “But if people don’t think innovation can be done safely, it will slow AI development and even keep us from moving forward. , protecting the foundations, and finally, explainability, one of the most important and most difficult technical problems in all of AI.”

On security, Schumer said, strengthening guardrails to prevent bad actors from using AI to harm others, and protecting people’s jobs to prevent mass unemployment and income inequality. is recommended. He believes companies that build and deploy software must take responsibility for its impact, including being audited.

But when we try to study models and algorithms internally, we need them to be explainable. Humans need to be able to understand how machines reach certain decisions. Schumer described this as the most difficult technical problem in AI.

“Now let me share my second proposal, a new legislative approach to move this framework into legislation. We plan to hold a series of AI Insights Forums “for AI policy,” he said.

IBM develops AI commentator for Wimbledon

The generative AI system built by IBM engineers will provide commentary at this year’s Wimbledon tennis tournament, which runs July 3-16.

The All England Club, which manages the Wimbledon venue, used Big Blue to develop new AI-powered features for Wimbledon’s official app and website. The system uses tools from the IBM watsonx platform to generate audio commentary and text captions summarizing the best parts of tennis matches captured in video highlights.

In its boilerplate, IBM said, “This tool is designed to give fans a more insightful experience as they catch up on key moments from the match with the Wimbledon app and highlight videos on wimbledon.com. It’s designed,” he said. “This year’s introduction is another step towards making commentary available in exciting ways outside of Wimbledon’s show court, where live human commentary is already available.”

AI commentators are trained on professional tennis jargon, but are not meant to replace them. “I don’t see AI as a replacement for the human element, but as a complement to it,” said Kevin Farrar, IBM Sports Partnerships Leader. telegraph paper.

“You can’t replace John McEnroe as commentator. The human element has to be there all the time. It’s very complementary and complementary. To provide commentary in the future for matches that do not: Senior, Junior, Wheelchair [events].”®