China and EU Strike Investment Deal: On December 30, Chinese and European Union leaders reached an agreement on a deal that would reduce investment barriers and give European companies more access to the Chinese market. Negotiations, which had been underway for nearly seven years, reportedly gained momentum when Germany, which has significant manufacturing interests in China, took over the rotating six-month presidency of the European Council last July. While the agreement will likely be a boon to European companies, a number of observers say the deal is a major strategic win for China. Representatives of the incoming Biden administration appeared to agree, publicly signaling their desire for the EU to delay the deal. Before the deal can take effect, it must be ratified by the European Parliament, where China’s human rights record could become a major barrier.
Stanford Medicine Blames Algorithm for Botched Vaccine Rollout: After its initial vaccine distribution plan excluded all but seven of its 1,300 medical residents, Stanford’s leaders blamed a flawed algorithm for the allotments and pledged to revise the rollout plan. What was described as a “very complex algorithm” by hospital leadership appears to have been a relatively simple rules-based formula that considered variables such as age, location and job-specific criteria. As vaccine distribution plans take shape nationwide, decision makers are turning to similar algorithms to help allocate scarce vaccine supplies. But experts note that even simple formulas like Stanford’s can produce bad results if the initial rules set by the algorithm’s designers are flawed. Observers worry these public failures and the lack of transparency that often accompanies decision-by-algorithm may undermine public trust and hinder the vaccine’s rollout.
- More: Stop Explaining Black Box Machine Learning Models for High Stakes Decisions and Use Interpretable Models Instead
U.S. Air Force Uses AI Aboard Military Plane for the First Time: Last month, the Air Force announced it had deployed AI on a military jet as part of a training flight, a first for the U.S. military. While a human pilot flew the U-2 high altitude reconnaissance aircraft, the AI system, dubbed ARTUµ, acted as the plane’s radar operator, controlling the plane’s sensor employment and tactical navigation. The ARTUµ system is based on the publicly available µZero algorithm developed by Alphabet subsidiary DeepMind to play games like chess and Go. The test was part of the Pentagon’s efforts to enable human-machine teaming, in which autonomous systems control peripheral tasks while humans retain control of core decisions like flight control and targeting.
Commerce Adds Dozens of Chinese Organizations to Export Blacklist: Last month, the Commerce Department added dozens of Chinese companies — including Semiconductor Manufacturing International Corporation, China’s largest semiconductor manufacturer — and several Chinese universities to the Entity List, restricting their access to U.S. exports unless the exporter receives government approval. The Commerce Department said exports of semiconductor manufacturing equipment used to produce advanced chips — 10 nanometers and below — to SMIC would be especially restricted, subject to a “presumption of denial” standard. All other export licenses would be handled on a case-by-case basis. Experts quickly took issue with this aspect of the rules, however, noting that the 10 nanometer restriction is narrowly drafted and would rarely apply in practice.
AI Funding in the 2021 Consolidated Appropriations Act: On December 27, President Trump signed the Consolidated Appropriations Act, 2021, which Congress passed a week earlier. It includes $2.3 trillion in funding for government operations and COVID-19 relief efforts. AI-relevant provisions include:
- An additional $5 million above the president’s request for JAIC operational systems development;
- An additional $5 million above the president’s request for foundational AI research;
- A requirement that JAIC inventory all AI programs at DOD; and
- An additional $65.5 million for Army AI research and development.
- Authorizes the Director of National Intelligence to award up to $15 million in contracts or grants to support microelectronics research; and
- Requires an assessment of critical trends in emerging technology, semiconductor supply chains, U.S. partner nation export controls and opportunities for greater export control collaboration in areas related to emerging technology.
- The National AI Initiative Act, which directs the president to establish a National Artificial Intelligence Initiative to support U.S. leadership in AI. The measure authorizes approximately $5 billion for National Science Foundation AI research and education grants, as well as more than $1 billion for Department of Energy AI research.
- Provisions from the CHIPS for America Act to support the development and production of semiconductors in the United States. The included provisions authorize financial incentives for the construction of semiconductor manufacturing facilities in the United States, direct DOD to establish a public-private consortium for the development of chips. They also authorize a fund to advance microelectronics research with allies and partners, provided those partners maintain export licensing policies on semiconductor technology exports to China equivalent to U.S. policies.
CSET’s translations of significant foreign language documents on AI
CSET’s translations of significant foreign language documents on AI
PRC Press Conference: Transcript of the National Development and Reform Commission October Press Conference. A translation of a short excerpt from the transcript of the October 20, 2020 PRC National Development and Reform Commission press conference. In response to a question from a reporter about failed projects to upgrade China’s microchip industry, the NDRC spokeswoman acknowledges that in the rush to develop the industry, many local governments in China have funded poorly thought-out projects led by incompetent companies. The spokeswoman proposes four measures to mitigate this problem.
Private Sector and PRC Defense S&T: CSET has translated five Chinese regulations governing the participation of private companies and private investors in the PRC defense science and technology industry, traditionally the exclusive domain of state-owned conglomerates. The regulations were issued between 2007 and 2014.
- Guide to the Participation of the Private Sector in Defense S&T (2007)
- Guiding Opinions on the Participation of the Private Sector in Defense S&T (2007)
- Opinions of the State Council on the Healthy Development of Private Investment (2010)
- Implementation Opinions on Channeling Private Capital into Defense S&T (2012)
- Opinions on Absorbing Private Companies into Weapons Research and Production (2014)
What We’re Reading
Report: Getting the future right – Artificial intelligence and fundamental rights, European Union Agency For Fundamental Rights (December 2020)
Blog Post: HAI Recommended Reading: 10 Books Worth Checking Out, Stanford University Human-Centered Artificial Intelligence (December 2020)
Article: Data and its (dis)contents: A survey of dataset development and use in machine learning research, Amandalynne Paullada, Inioluwa Deborah Raji, Emily M. Bender, Emily Denton and Alex Hanna (December 2020)
What’s New at CSET
- Mapping U.S. Multinationals’ Global AI R&D Activity by Roxanne Heston and Remco Zwetsloot
CSET is hiring! Please share with qualified candidates in your network or consider applying.
- Research Analyst, closes Friday (Jan 8): We’re looking for a Research Analyst to focus on biotechnology efforts. Collaborate with Research Fellows to develop and execute research projects. BA in relevant area and comfort with data analysis/viz required (MA preferred).
- FedScoop: Saving the federal cyber and AI workforce from obsolescence: How to attract and retain a new generation by Cindy Martinez
- Defense One: US Tech Firms Must Stop Helping China’s Defense-Linked Organizations by Ryan Fedasiuk and Emily Weinstein
- Military Operations Research: Developing an Analytical Method for Assessing Readiness, Risks, and Opportunity by Igor Mikolic-Torreira
CSET has launched a crowd forecasting platform. Sign up as a forecaster, and take a look at some of the predictions so far:
- (New) How many members will the Alphabet Workers Union have by December 31, 2021?
- (New) How many AI papers will be posted on arXiv between July 1 and December 31, 2021, inclusive?
- (New) What percentage of U.S. news articles about AI will be about privacy and security between July 1 and December 31, 2021, inclusive?
- American Public Media: Anna Puglisi appeared on Marketplace to discuss the attempted theft of U.S. semiconductor secrets by China.
- NPR: WBUR’s Here & Now brought on CSET’s Ben Buchanan to talk about the recent SolarWinds cyberattack against the U.S. government.
- Associated Press: Ben Buchanan was also quoted by the AP in an article about the cyber operation.
- South China Morning Post: Will Hunt spoke with the South China Morning Post about the Biden administration’s approach to semiconductor exports.
- January 11-15 and 19-21: AIAA, 2021 AIAA SciTech Forum featuring Melissa Flagg
- January 14: CSET, The Way Forward: Tech Policy Recommendations for the Biden Administration featuring Helen Toner, Andrew Imbrie, Melissa Flagg and Remco Zwetsloot
What else is going on? Suggest stories, documents to translate & upcoming events here.