DeepMind Achieves Major Success: In what experts have heralded as “a breakthrough of the first order,” DeepMind, the UK-based AI company owned by Alphabet Inc., successfully modeled protein structures with an accuracy equivalent to much more expensive and time-consuming methods. Proteins are notoriously difficult to map, but knowing their structures can help researchers understand diseases, develop medicines, make biofuels, and more. DeepMind’s AlphaFold program beat out dozens of other teams at this week’s biennial “Critical Assessment of Structure Prediction” challenge, modeling proteins for which the structures are known but not yet public. The program slightly outperformed current experimental methods overall, and nearly matched current methods against the most challenging proteins. As a condition of entering the CASP competition, all groups agree to reveal information about their methods sufficient to replicate the process. Demis Hassabis, DeepMind’s CEO and co-founder, said the company was looking into sharing the technology with other researchers.
- More: DeepMind’s protein-folding AI has solved a 50-year-old grand challenge of biology | 2017: Google’s AlphaGo Defeats Chinese Go Master in Win for A.I. | 2018: Google’s DeepMind aces protein folding
- More: Despite past denials, LAPD has used facial recognition software 30,000 times in last decade | Face for sale: Leaks and lawsuits blight Russia facial recognition
DOD Inspector General Highlights Tech Challenges: In mid-November, the Pentagon’s Office of Inspector General released its annual report summarizing the biggest management and performance challenges facing the agency in the coming fiscal year. It identified 10 challenges, including “Building and Sustaining the DoD’s Technological Dominance,” “Securing the DoD’s Information Systems, Networks, and Data” and “Transforming Data Into a Strategic Asset.” The report highlighted the potential of artificial intelligence and machine learning to “revolutionize how war is conducted,” but stressed the need to maintain “realistic expectations” about AI/ML’s integration and future use. It also emphasized the importance of protecting technological advantages, citing inconsistent security practices among DOD contractors. The report’s section on “transforming data into a strategic asset” and its emphasis on “cultivating a culture that understands how to use data” echoed the conclusions of the Pentagon’s Data Strategy published earlier this year.
OMB Releases Guidance on AI Regulation: On November 17, the Office of Management and Budget issued guidance on regulating artificial intelligence applications, as directed by a 2019 executive order. Pursuant to that directive to “consider ways to reduce barriers to the use of AI technologies in order to promote their innovative application,” the OMB memo says federal agencies “must avoid regulatory or non-regulatory actions that needlessly hamper AI innovation and growth,” and “[w]here AI entails risk, agencies should consider the potential benefits and costs of employing AI, as compared to the systems AI has been designed to complement or replace.” AI-regulating agencies are now expected to review the memorandum, submit plans to OMB detailing how they plan to align with its directives, and provide lists and descriptions of planned or considered AI regulations.
NSTC Releases Two Reports: Two National Science and Technology Council subcommittees published reports relating to artificial intelligence and emerging technologies:
- Pioneering the Future Advanced Computing Ecosystem: A Strategic Plan, by the Subcommittee on Future Advanced Computing Ecosystem, lays out a plan for a nationwide computing ecosystem capable of driving U.S. research and economic competitiveness.
- Recommendations for Leveraging Cloud Computing Resources for Federally Funded Artificial Intelligence Research and Development, by the Select Committee on Artificial Intelligence, makes recommendations for how to make use of cloud computing in federally funded AI R&D, such as launching pilot projects using commercially available cloud services to conduct publicly funded research.
CSET’s translations of significant foreign language documents on AI
CSET’s translations of significant foreign language documents on AI
China’s 14th Five-Year Plan: Proposal of the CCP Central Committee on the 14th Five-Year Plan for National Economic Development. The 14th Five-Year Plan Proposal, approved at the Chinese Communist Party plenum in October, sets near-term priorities for PRC economic development, including for China’s tech sector and emerging industries.
PRC Ministry of Science and Technology: Chinese R&D Analysis of the State of Chinese R&D Personnel Development in 2018. China’s research and development personnel pool is the largest in the world and growing rapidly, according to this statistical report by the PRC Ministry of Science and Technology. However, the report also notes that on a per capita basis, China has fewer R&D workers than the West.
What We’re Reading
Report: 2020 Annual Report to Congress, U.S.-China Economic and Security Review Commission (December 2020)
Paper: To Compete, Invest in People: Retaining the U.S. Defense Enterprise’s Technical Workforce, Morgan Dwyer, Lindsey R. Sheppard, Angelino Hidalgo and Melissa Dalton, CSIS (November 2020)
Article: China’s President Xi Jinping Personally Scuttled Jack Ma’s Ant IPO, Jing Yang and Lingling Wei, The Wall Street Journal (November 2020)
Paper: Preventing Repeated Real World AI Failures by Cataloging Incidents: The AI Incident Database, Sean McGregor (November 2020)
What’s New at CSET
- Automating Cyber Attacks: Hype and Reality by Ben Buchanan, John Bansemer, Dakota Cary, Jack Lucas and Micah Musser
- U.S. Demand for Talent at the Intersection of AI and Cybersecurity by Cindy Martinez and Micah Musser
- CSET: Chinese Talent Program Tracker by Emily Weinstein
- Defense One: Mapping China’s Sprawling Efforts to Recruit Scientists by Emily Weinstein
- Defense One: Are AI Professionals Actually Unwilling to Work for the Pentagon? by Catherine Aiken and Margarita Konaev
- ChinaFile: How Can the U.S. Maintain Global Leadership in Science and Technology? with Helen Toner
- OECD.AI: A first look at the OECD’s Framework for the Classification of AI Systems, designed to give policymakers clarity with Dewey Murdick
- Taiwan Insight: Technology Without Authoritarian Characteristics: An Assessment of the Taiwan Model of Combating COVID-19 by Emily Weinstein
- CSET: CSET Experts in the News
CSET has launched a crowd forecasting platform. Sign up as a forecaster, and take a look at some of the predictions so far:
- (New) Will the United States re-enter the 2015 nuclear deal with Iran by December 31, 2021?
- (New) When will Ant Financial have an IPO in China or Hong Kong?
- (New) What will be the value, in dollars, of all Chinese imports of semiconductor chips in 2021?
- (New) What will be the value, in dollars, of all Chinese imports of semiconductor manufacturing equipment in 2021?
- FedScoop: The recent issue brief by Catherine Aiken, Rebecca Kagan and Michael Page, “‘Cool Projects’ or ‘Expanding the Efficiency of the Murderous American War Machine?’: AI Professionals’ Views on Working With the Department of Defense” was recapped by FedScoop.
- Axios: An Axios article about the future of autonomous weapons in the military referenced Margarita Konaev’s recent three-part series “U.S. Military Investments in Autonomy and AI: Costs, Benefits, and Strategic Effects.”
- Axios: Axios also summarized the results of the CSET brief “‘Cool Projects’ or ‘Expanding the Efficiency of the Murderous American War Machine?’: AI Professionals’ Views on Working With the Department of Defense.”
- The New York Times: CSET Founding Director Jason Matheny was quoted in a New York Times piece about U.S. technology being used to power the Chinese surveillance state.
- Axios: An Axios story about the lack of U.S.-China cooperation in space cited two papers by CSET authors — one by Senior Fellow Matthew Daniels and the other by Visiting Researcher Lorand Laskai with Richard Danzig.
- December 8: CSET, Following the Money: What Investment Trends Can Tell Us About AI’s Present and Future featuring Zachary Arnold, moderated by Sarah Sewall
- December 15: CSET, Forecasting U.S.-China [De]Coupling in the Semiconductor Industry featuring Saif M. Khan, moderated by Michael Page
- January 11-15 and 19-21: AIAA, 2021 AIAA SciTech Forum featuring Melissa Flagg
What else is going on? Suggest stories, documents to translate & upcoming events here.