IBM wins the 2nm chip race, SecDef Austin links AI and deterrence, and Tesla pumps the brakes on autonomous driving

Sign up
Worth Knowing

IBM Designs First 2nm Chip: Last week, IBM announced it had developed the world’s first 2 nanometer semiconductor. While the advance promises significant performance and efficiency improvements over today’s 7nm chips — IBM projects 45 percent higher performance or 75 percent lower energy consumption — it will probably be several years until chips designed using the process reach the market. While the company produced the world’s first 7nm chip in 2015, it wasn’t until late 2018 that they began to reach consumers. In the meantime, chips from slightly less-advanced process nodes are on the way: Samsung and TSMC began high-volume production of 5nm chips last year, and 3nm chips are expected to enter production in 2022. But IBM’s 2nm process is an important proof of concept at a time when AI systems require more speed (and energy) than ever.
Where’s the AI?: In the last edition of policy.ai, we discussed the European politicians who said they’d been hoodwinked by someone using deepfake technology to impersonate Leonid Volkov, a Russian opposition leader. As The Verge reported soon after, the prank appears to have been much less high tech: Using makeup, dim lighting and a low-quality camera, the Russian pranksters say they managed to pull off a passing Volkov impression without the aid of algorithmic trickery. This isn’t the first time a seemingly incorrect deepfake accusation has been at the center of a public story. In 2018, for instance, many speculated that a video of Gabon’s president was a deepfake meant to cover up serious health problems. While experts never turned up evidence that the video had been altered, that didn’t stop the rumors from spreading widely online. As observers noted at the time, actual AI alteration isn’t necessary to sow confusion — just the existence of the technology is enough to create doubt and set off rumors.
Self-Driving Cars? Not This Year, Tesla Tells Regulators: Back in January, Tesla CEO Elon Musk told investors his company would achieve autonomy “in excess of human” capabilities by the end of the year, but a newly-released memo to California regulators threw cold water on that dream. Musk’s statement “does not match engineering reality,” Tesla’s director of Autopilot software told the California DMV — in fact, the company’s driver-assist features are still only level 2 autonomous, requiring constant supervision from the driver. That’s a far cry from level 5 autonomy — the highest level, requiring no input from a driver under any conditions. After a spate of crashes in which drivers were using or may have been using Tesla’s Autopilot system, the company is facing criticism from those who say names like “Autopilot” and “Full Self-Driving” (two of Tesla’s driver-assistance features) lead some consumers to believe their cars are actually autonomous. While the cars are not approved for driverless operation, that has not stopped some drivers from texting, playing video games, or even sleeping behind the wheels of their Teslas.   
Government Updates

SecDef Austin Highlights Importance of Emerging Tech: In his first major speech as Secretary of Defense, Lloyd Austin emphasized the central role emerging technologies — including AI — will play in maintaining deterrence against potential adversaries. Austin’s remarks came during the change of command ceremony for the U.S. Indo-Pacific Command, the combatant command responsible for both China and India. After a brief introduction, he quickly turned to the impacts of new technologies. “The way we’ll fight the next major war is going to look very different from the way we fought the last one,” Austin said. “In this young century, we need to understand faster, decide faster, and act faster. Our new computing power isn’t an academic exercise. This is about real-world, real-time advantages — about fully grasping a situation moving at the speed of war.” The secretary’s message was echoed later that day by Deputy Secretary Kathleen Hicks and adapted into an op-ed published in The Washington Post last week.    

The White House Launches AI Website: The White House launched a new website last week, AI.gov, that aims to make information on AI more accessible to the general public. The site is operated by the National Artificial Intelligence Initiative Office, the creation of which was mandated by the National AI Initiative Act (we covered the NAII’s establishment in January). The site collects information on the AI efforts of many federal agencies and programs, including the NSF, the State Department and DOD. It also serves as a central repository for many of the federal government’s AI-relevant documents, including recent legislation, executive orders and strategy documents. Lynne Parker, the NAII’s founding director, told Axios that the website will help researchers access “both the computing and the data they need in order to do cutting edge research,” but it is not yet clear how it will do that. At the moment, the site links to AI-relevant data and educational and career resources from other programs and agencies, but Parker’s quote may indicate NAII-specific resources are on the way.

Endless Frontier and Safeguarding American Innovation Acts on the Move: The Senate Commerce Committee approved an amended version of the Endless Frontier Act yesterday by a vote of 24 to 4, following a six-hour markup for which more than 200 amendments were filed. The bill would establish a new technology directorate within the National Science Foundation and direct $100 billion in new funding to the NSF. Meanwhile, the Senate Homeland Security and Governmental Affairs Committee favorably reported the Safeguarding American Innovation Act by voice vote. This bipartisan proposal from Sens. Portman and Carper seeks to address threats related to foreign talent programs and intellectual property theft. Following the markup, Senator Portman said his intent is for the bill to be included in the Endless Frontier Act during full Senate consideration. Various Senators have indicated plans for the full Senate to consider the Endless Frontier Act by the end of May.

New “Data Decrees” Aim to Make Pentagon a Data-Centric Organization: On May 5, Deputy Secretary of Defense Kathleen Hicks issued a memo introducing five “Data Decrees” that aim to improve DOD’s handling of data across the department. The decrees emphasize the importance of data centralization, maximizing sharing, interoperability, safe storage, and adherence to industry best practices. Last year, the DOD introduced a new data strategy that touched on many of the same themes (we covered the strategy’s release). While that document was produced under the previous administration, Hicks’ memo indicates the Biden DOD sees the issue of data prioritization in a similar light.

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

China’s 14th Five-Year Plan: Outline of the People’s Republic of China 14th Five-Year Plan for National Economic and Social Development and Long-Range Objectives for 2035. This document is China’s 14th Five-Year Plan, covering 2021-2025, as passed by the Chinese parliament, the National People’s Congress, in March 2021. Although the Five-Year Plan contains relatively few quantitative targets, it details a vast array of near-term PRC economic, trade, S&T, defense, political, social, cultural, environmental, and other policy priorities. The 14th Five-Year Plan differs from past plans in that it also includes a short section on “long-range objectives” for 2035. Note that although this document is an “outline,” the PRC government has labeled the longest and most authoritative full versions of previous five-year plans it released as “outlines” as well.

PRC Cybersecurity Competitions: Notice on Regulating the Promotion of Cybersecurity Competitions. This regulation, issued in 2018, requires personnel involved in major PRC government or military projects to obtain police approval before participating in overseas cybersecurity competitions. The regulation also compels competitors to disclose national security- and stability-related vulnerabilities they discover during competitions to the police.

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: The Department of Defense’s Looming AI Winter, Marc Losito and John Anderson, War on the Rocks (May 2021)

Article: US-China tech war: Beijing’s secret chipmaking champions, Cheng Ting-Fang and Lauly Li, Nikkei (May 2021)

Article: The Chip Shortage Keeps Getting Worse. Why Can’t We Just Make More?, Ian King, Adrian Leung and Demetrios Pogkas, Bloomberg (May, 2021)

What’s New at CSET

JOB OPENINGS
Please share with qualified candidates in your network or consider applying:
  • Policy Communications Analyst: CSET is currently seeking applications for a Policy Communications Analyst to help advance our written and visual products by advising researchers on report clarity, narrative flow, cogency, presentation effectiveness and parsimony.
  • Georgetown’s Walsh School of Foreign Service, the home institution for CSET, is hiring for the Gracias Family Chair in Security and Emerging Technology. This two-year, non-tenure-track, rank-open position will have teaching and administrative responsibilities in the Security Studies Program, with potential for affiliations with CSET and the Science, Technology and International Affairs Program. The successful candidate will have experience focused on security and emerging technology, with a particular focus on AI and its implications for national and international security. Applications received by May 16 will receive full consideration.
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
Upcoming Events


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