DOJ files antitrust case against Google, AMD buys Xilinx, China passes export control law, and NSCAI recommends 66 reforms

Sign up
Worth Knowing

China Moves Forward With Export Control and Data Protection Laws: Beijing passed a new export control law on October 17 after months of U.S. regulatory pressure on Chinese firms such as Huawei and ByteDance. There is no published list of controlled items; however, the text of the law, which is to take effect December 1, says it will apply to “goods, technology, or services related to … the protection of national security.” Chinese state media indicated it could be used to restrict the export of rare-earth metals needed for production of U.S. high-end technology. Meanwhile, Beijing published a draft of a new personal data protection law, a first of its kind for China. Experts say the long-anticipated law, introduced after a related Data Security Law was released for public comment in July, “marks a significant evolution in China’s data protection regime.” The draft text says violators could be fined up to 50 million yuan or 5 percent of annual revenue. 
AMD Buys Chip Company Xilinx for $35B: Yesterday, U.S. semiconductor company AMD announced plans to purchase fellow Silicon Valley chip designer Xilinx for $35 billion in stock. The deal brings in Xilinx’s experience with field programmable gate arrays, a type of chip used in artificial intelligence, communications, and data center applications. Observers noted the similarities between AMD’s deal and acquisitions made by its chief rivals, Intel and Nvidia. Intel’s 2015 purchase of FPGA designer Altera made it a leader in communications chip design, while Nvidia’s $40 billion deal to acquire British chipmaker Arm, announced last month, indicated plans to break into the CPU market. AMD and Xilinx said they expected the acquisition to be finalized by the end of 2021, but the deal still needs shareholder and regulatory approval, which could be a major hurdle. Nvidia’s purchase of Arm has already come under scrutiny from regulators in Beijing, putting that deal in jeopardy.
Businesses, Trade Groups and Colleges File Suit Over H-1B Changes: A group of trade organizations, businesses and colleges — including the U.S. Chamber of Commerce, Cornell University and Caltech — sued the Departments of Labor and Homeland Security after changes to the H-1B visa program earlier this month. Those reforms would exclude an estimated third of current recipients by reducing the list of qualifying occupations, tightening degree requirements, and significantly raising minimum salaries. Filed in the U.S. District Court for the Northern District of California, the complaint alleges the new rules “would sever the employment relationship of hundreds of thousands of existing employees in the United States.” Earlier this month, the same U.S. District Court temporarily blocked the Trump administration’s June order to suspend H-1B processing.
Machine Learning Spotlight — “Less Than One”-Shot Learning: A new paper from the University of Waterloo in Ontario proposes a way to get AI models to identify more objects than the number of examples used to train them. The paper builds on a previous method used by MIT researchers, who reduced a popular dataset used to train AI models to recognize written numerals from 60,000 images down to 10. Despite using a much smaller dataset, the MIT model was nearly as accurate as those trained on the larger dataset. The Waterloo researchers managed to reduce the dataset even further, using five images to train their model. Because the cost of the computing power necessary to run massive datasets can be exorbitant, this research — and research on “Few-Shot” and “One-Shot” learning — could help open up machine learning development to smaller firms and research programs.
Government Updates

U.S. Files Antitrust Case Against Google: After more than a year of investigation, the Justice Department filed an antitrust lawsuit against Google last week. The complaint accuses the company of “unlawfully maintaining monopolies in the markets for general search services, search advertising, and general search text advertising.” DOJ’s case centers on Google’s practice of negotiating deals with smartphone makers and web browser developers to make Google the default search engine on their products. While breaking up the company is one possible outcome of the case, that result is thought to be unlikely. DOJ’s lawsuit could take years to resolve and may receive more support soon. In addition to the 11 states that signed on to the DOJ case, attorneys general from seven other states are conducting parallel investigations and have stated they may join the suit soon.

NSCAI Releases Interim Report and Recommendations: On October 13, the National Security Commission on Artificial Intelligence submitted its 2020 Interim Report and Third Quarter Recommendations. Its 66 recommendations span three areas: preparing for international AI and emerging technology competition, expanding the domestic AI talent pipeline, and working with allies to build and sustain international AI cooperation. The commission proposed an Executive Order to advance AI technical standards through improved interagency coordination, new funding for NIST and collaboration with industry and allies. Additionally, the commission recommended new mechanisms for investing in AI talent, such as flexible awards for promising individuals and teams.

White House Releases Strategy for Critical and Emerging Technologies: On October 15, the White House published its National Strategy for Critical and Emerging Technologies. Developed by the National Security Council, the strategy is framed as a response to Russia and China’s “state-directed models” of R&D. Its “market-oriented approach” is built around two “Pillars of Success.” The first, “Promote the National Security Base,” calls for developing the science and technology workforce and adopting “innovation-friendly regulations.” The second, “Protect Technology Advantage,” aims to deter intellectual property theft by foreign competitors. Observers noted that the high-level document contained few specific details and that the timing of the strategy’s release —  less than three weeks before an election — may mean it gets overlooked.

In Translation
CSET’s translations of significant foreign language documents on AI

Science and Technology Spending Overview: 2019 Nationwide Statistical Communiqué on Science and Technology Spending and Investment. This document, issued by three PRC ministries, provides statistics on Chinese technical R&D spending in 2019, broken down by industry and region. Although China’s R&D spending increased significantly across the board in 2019, product development — as opposed to basic or applied research — still accounts for more than 80% of it.

What We’re Reading

Paper: Emerging Technologies and Trade Controls: A Sectoral Composition Approach, Lindsay Rand, Andrea Viski and Tucker Boyce (October 2020)

Paper: Charting a Transatlantic Course to Address China, Center for New American Security (October 2020)

Paper: Forecast 2025: China Adjusts Course, Macro Polo (October 2020)

What’s New at CSET

REPORTSJOB OPENINGS
CSET is hiring! Please share with qualified candidates in your network or consider applying. Applications close on November 1.
  • Research Analyst: Collaborate with Research Fellows on research projects about workforce and education related issues. BA in relevant area and comfort with data analysis/viz required.
PUBLICATIONS AND PODCASTSFORECASTSEVENTS
  • On October 19, CSET hosted a webinar, Using Crowd Forecasting to Inform Policy, featuring Founding Director Jason Matheny and moderated by Research Fellow Michael Page. The event coincided with the release of a new issue brief, Future Indices, by Page, Catherine Aiken and Dewey Murdick, which discusses how Foretell can inform policy by providing data on future trends.
IN THE NEWS
Upcoming Events


What else is going on? Suggest stories, documents to translate & upcoming events here.