Intel’s $20bn U.S. chipmaking investment, AI flops on COVID-19 screening, and DARPA’s virtual dogfights show promise

Sign up
Worth Knowing

Intel to Make Major Investment in U.S. Chipmaking: Last week, Intel Corporation announced plans to spend $20 billion on two new semiconductor manufacturing facilities (fabs) in Arizona. The company also unveiled plans to establish a standalone foundry business, Intel Foundry Services, that will provide U.S. and Europe-based chipmaking capacity to third-party companies. While Intel confirmed that it will outsource some of its other chipmaking to external foundries (which we discussed earlier this year), these major commitments to domestic manufacturing will likely help allay fears that its outsourcing plans would doom U.S. chipmaking. Intel is self-funding the Arizona investment, but a company official said it would look to secure subsidies from the Biden administration and foreign governments as it continues to expand its manufacturing capacity in the United States and Europe. The announcement comes as the global economy continues to suffer from a persistent chip shortage, but it will do little to ease the crunch — Intel’s new fabs are not expected to come online until 2024. 
Researchers Find Flaws in Machine Learning-Based COVID-19 Detection: A review of more than 300 machine learning models meant to help COVID-19 detection and prognostication concluded that none of the models met the standard for clinical use. The study, published in the journal Nature Machine Intelligence, found that none of the models were clinically useful and that most suffered from one or a handful of common problems, such as lack of external data validation, poor reproducibility and biases in study design. The authors recommend that, in future, researchers and reviewers should assess models against checklists — including the recently developed Checklist for Artificial Intelligence in Medical Imaging — to avoid these pitfalls.  
Human Rights Group Warns of AI Abuse in Myanmar: Human Rights Watch warned that Myanmar’s military’s access to artificial intelligence poses “a serious threat to basic rights in the country.” In December last year, Myanmar installed 335 Huawei surveillance cameras, each equipped with facial recognition and license plate scanning capabilities, in the capital city of Naypyidaw. Following the February 1 military coup, control of those cameras and access to their data is now presumably in the hands of security forces that have reportedly killed more than 500 pro-democracy protesters. Observers note that the military has also gained access to advanced tools from Western countries — such as European iPhone-cracking devices and U.S. hacking software — but access to Huawei cameras has raised specific concerns because of how they have been used in China, namely to monitor and track Uyghurs in China’s Xinjiang region.
Government Updates

House Science Committee Leaders Introduce Bill to Reshape NSF: A bipartisan group of representatives, led by Chairwoman Johnson of the House Committee on Science, Space, & Technology introduced the National Science Foundation for the Future Act, which would double NSF’s budget over the next five years. It would also establish a new “Directorate for Science and Engineering Solutions” within NSF with an annual budget that would grow to $5 billion over five years. The directorate would focus on “societal and national challenges” including climate change, cybersecurity, social and economic inequality, and workforce development, among others. House Science Committee staff compared it with the Endless Frontier Act, reportedly saying that bill, introduced by Sens. Schumer and Young last year, was too focused on U.S.-China competition and set unrealistic funding targets. Staff noted that the proposal could be packaged with other related legislation in the future, but specific plans for moving the bill have not been announced.

DARPA Details Progress on Autonomous Air Combat Project: The Pentagon’s Defense Advanced Research Projects Agency announced that it recently tested teams of AI-controlled virtual F-16 fighter jets in simulated dog fights against an autonomous enemy plane. Conducted in February as part of DARPA’s Air Combat Evolution (ACE) program, the 2-v-1 simulated engagements represent another significant step for a program still in its first phase; in August of last year, an AI-controlled F-16 beat a human pilot 5-0 in a virtual dogfight. The success of February’s tests brings DARPA closer to testing AI-controlled planes in real world conditions; DARPA says it is now preparing for ACE Phase 2, planned to begin later this year, during which it will test its AI in live dogfights involving subscale aircraft. If those tests are successful, Phase 3 will escalate to full-scale jets in live-fly team dogfights in 2023 and 2024.

House Science Committee Staff Warn of Federal Brain Drain: The House Science Committee’s majority staff released a report detailing worrying workforce trends at several federal science agencies. According to the report, which was prepared for a related subcommittee hearing last month, the overall workforce of the seven science agencies within the committee’s jurisdiction fell 6.5% between FY2009 and FY2020. DOE, EPA and NOAA saw even larger declines, while headcounts at the other four agencies — NASA, NIST, NSF and DHS S&T — either remained steady or increased. To shore up the workforce and address persistent racial, ethnic and gender employment gaps in the federal science agencies, the report’s authors recommend increasing funding for the relevant agencies, committing to diversity in the federal STEM workforce and stepping up recruitment efforts.

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

Russian AI Subsidies: Draft Resolutions have been Prepared Concerning the Allocation of 16.5 Billion Rubles in Subsidies for the Development of AI in Russia. These are draft resolutions indicating where the Russian government plans to invest U.S. $217 million in AI development from 2021 to 2023. The Ministry of Economic Development intends to disburse funding to several nonprofit organizations that will construct platforms for national AI pilot projects, fund small enterprises working on AI, and run AI-based hackathons using datasets held by the Russian government.

Russian AI Notice: Development of Artificial Intelligence. This notice establishes the Ministry of Economic Development (MoED) as the main AI policymaking body of the Russian government. It defines what “AI” means to Russia, and describes the Russian government’s plan to develop a robust AI industry. The plan stipulates that the Russian government will promote funding for science and technology, acquire datasets and hardware needed for AI development, formulate new methods to regulate AI, and absorb foreign capital and knowledge to accelerate the development of a Russian AI industry. Finally, the notice announces the creation of an AI subcommittee co-chaired by MoED Minister Maxim Reshetnikov and Sberbank Chairman Herman Gref.

Xi on China’s S&T Power: Xi Jinping: ‘Strive to Become the World’s Primary Center for Science and High Ground for Innovation’. This speech by PRC President Xi Jinping describes China’s ongoing efforts to catch up to the United States and other world leaders in key technologies and to improve China’s capacity for S&T innovation. This translation is provided by the New America-Stanford University DigiChina Project and includes translation contributions by CSET staff.

If you have a foreign-language document related to security and emerging technologies that you’d like translated into English, CSET may be able to help! Click here for details.

What We’re Reading

Article: Young, deported, and learning to code, Caroline Tracey, Rest of World (March 2021)

Article: Artificial Intelligence, Chatbots Threaten Call-Center Industry, Human Operators, Bruce Einhorn, Siegfrid Alegado and Ditas B Lopez, Bloomberg (March 2021)  

Report: Degrees of Separation: A Targeted Approach to U.S.-China Decoupling – Interim Report, Stephanie Segal and Dylan Gerstel, CSIS (February 2021)

Report: Building an AI World: Report on National and Regional AI Strategies, Second Edition, Johnny Kung, CIFAR (May 2020)

What’s New at CSET

REPORTSPUBLICATIONS AND EXPERT TESTIMONYFORECASTS
CSET maintains a crowd forecasting platform. Sign up as a forecaster, and take a look at some of the predictions so far:
EVENTSIN THE NEWS
Upcoming Events


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