Google AI designs better chips than humans do in a fraction of the time, the U.S. and EU team up on tech, and the White House launches an AI task force

Sign up
Worth Knowing

Better Chips, Faster — Google Says Its AI-Powered Chip Designer Beats Humans: The next generation of Google’s specialty AI chips — Tensor Processing Units — was designed by a new AI system that can map out superior chips in a fraction of the time it takes humans. “Floorplanning” — the process of laying out the components of a microchip — can take experts months of work. Google’s new system, as described in an article in the journal Nature, can create a floorplan that matches or exceeds human designs in key metrics such as performance and power consumption in under six hours. The system’s designers analogized floorplanning to a complex board game and used reinforcement learning — the same method the Google subsidiary DeepMind used to master games such as Go and StarCraft — to train it. The system’s designers say it could lead to a symbiotic relationship in which AI-designed chips run the systems that create more powerful AI chips, accelerating both AI and chip development.
China Claims Its AI Can Beat Human Pilots in Battle: Chinese state media reported that an AI system had successfully defeated human pilots during simulated dogfights. According to the Global Times report, the system had shot down several PLA pilots during a handful of virtual exercises in recent years. Observers outside China noted that while reports coming out of state-controlled media outlets should be taken with a grain of salt, the capabilities described in the report are not outside the realm of possibility. Last year, for example, an AI agent defeated a U.S. Air Force F-16 pilot five times out of five as part of DARPA’s AlphaDogfight Trial (which we covered at the time). While the Global Times report indicated plans to incorporate AI into future fighter planes, it is not clear how far away the system is from real-world testing. At the moment, the system appears to be used only for training human pilots. DARPA, for its part, is aiming to test dogfights with AI-piloted subscale jets later this year and with full-scale jets in 2023 and 2024.
European Privacy Developments: Two pieces of news out of the European Union foreshadow potential developments in privacy regulation:
Government Updates

United States and European Union Form Transatlantic Tech Council: Last week, the United States and the European Union announced the creation of the U.S.-EU Trade and Technology Council (TTC), a joint effort to reduce trade barriers, align regulatory standards and counter China. According to the statement released after the joint U.S.-EU summit in Brussels, the TTC will include 10 working groups focusing on issues such as standards cooperation (including on AI and other emerging technologies), data governance, investment screening, and strengthening critical supply chains. The TTC was initially proposed by the European Commission late last year as part of a broader push for transatlantic cooperation in the wake of President Biden’s election (see our coverage of the proposal here). The TTC will reportedly be co-chaired by Secretary of State Antony Blinken, Secretary of Commerce Gina Raimondo and U.S. Trade Representative Katherine Tai.  

White House Launches AI Task Force: The National Science Foundation and the Office of Science and Technology Policy have announced the launch of the AI Research Resource Task Force. Comprising a dozen technical experts from government, private companies and academia, the task force will create a roadmap for the National AI Research Resource: “a shared research infrastructure” meant to provide computing power, open government and non-government datasets, and educational tools to students and AI researchers, according to the White House’s announcement. Both the task force and the NAIRR were mandated as part of the National AI Initiative Act, which was passed as part of the FY2021 NDAA. That act also established the National AI Initiative Office, whose director, Lynne Parker, will co-chair the task force. The NAIRR task force will deliver an interim report to Congress in May 2022 and a final report in November 2022.  

Pentagon Announces Data Readiness Initiative: On Tuesday, Deputy Secretary of Defense Kathleen Hicks announced a new DOD program — the AI and Data Acceleration initiative — that aims to prepare combatant commands for greater AI integration. As part of the initiative, “operational data teams” will be sent to each of the 11 combatant commands to prepare their data to be “AI-ready.” “Flyaway teams of technical experts” will also help the combatant commands integrate AI to improve and automate workflows. Last month, Defense Secretary Lloyd Austin reportedly approved the DOD’s Joint All-Domain Command and Control (JADC2) strategy — an effort to interconnect its weapons systems and sensors and incorporate emerging technologies like AI. While the strategy remains classified, the initiative announced by Hicks this week appears to be a major step toward its realization.

House Moves Ahead With Its National Science Foundation Bill: Last week, the House Science Committee approved the NSF for the Future Act by voice vote, advancing its alternative to the Endless Frontier Act, which the Senate had passed the week prior as part of its U.S. Innovation and Competition Act (USICA). As we noted in the last edition of policy.ai, the Senate bill would pour more than $200 billion into U.S. R&D and high tech manufacturing — including $81 billion to the NSF, $29 billion of which would go to a new directorate focused on emerging technology. The House’s proposal includes $72 billion for the NSF overall and a significantly smaller directorate. Next steps for the bill are unclear. House Foreign Affairs Chairman Gregory Meeks introduced the EAGLE Act “in response to the policy challenges posed by the People’s Republic of China.” The House could package these and other measures together, similar to the Senate’s USICA, but formal plans have not been announced. For more on the NSF for the Future Act and other legislative news, see our latest Legislative Roundup.

CSET Job Openings

Please share with qualified candidates in your network or consider applying:
  • Research Fellow – AI TEV&V will focus on the safety and risk of deployed AI systems by researching real-world AI incidents and use these identified incidents with other analyses of AI systems to inform policy recommendations regarding AI safety, test, evaluation, verification and validation (TEV&V) processes, standards setting and management, and the appropriate employment and operation of AI systems by businesses and the US Government (including the military). Applications due by July 19
In Translation
CSET’s translations of significant foreign language documents on AI

PRC Semiconductor Plan: Application Guidelines for the “Chips, Software, and Compute” (Chip Category) Major Special Project of the 2020 Annual Guangdong Provincial Program for Research and Development in Key Fields 2020. This document is an example of a PRC provincial government plan to boost the semiconductor manufacturing industry. The highly granular and technical plan aims to elevate the quality of Guangdong Province’s general purpose and specialized microchips to the leading level in China, and in some cases, to the global cutting edge.

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’s New at CSET

REPORTSPUBLICATIONSFORECASTS
CSET maintains a crowd forecasting platform. Sign up as a forecaster, and take a look at some of the predictions so far:
IN THE NEWS
What We’re Reading

Article: Reward is enough, David Silver, Satinder Singh, Doina Precup and Richard S. Sutton, Artificial Intelligence (May 2021) 

Article: What Really Happened When Google Ousted Timnit Gebru, Tom Simonite, Wired (June 2021) 

Report: Funding Risky Research, Chiara Franzoni, Paula Stephan and Reinhilde Veugelers, NBER (June 2021)   

Upcoming Events


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