|
Wednesday, November 11, 2020
|
|
by ALEX FRIEDLAND
|
|
Worth Knowing
U.S. Election Recap — Tech Initiatives That Passed and Failed: While the November 3 elections had most eyes focused on outcomes at the federal level (see “Government Updates” below), several state- and local-level ballot initiatives had tech policy implications:
-
California voters struck down a referendum that would have replaced cash bail with an algorithmic risk assessment. Lawmakers had approved the algorithmic system in 2018; courts delayed its intended October 2019 implementation pending this year’s vote. Supported by critics of cash bail and California Democratic Party leaders, the law was opposed by the California Republican Party, Human Rights Watch, the ACLU and the NAACP due, in part, to fears that an algorithmic assessment could entrench racial biases.
-
Michigan voters approved Proposal 2, an amendment to the state constitution that requires search warrants to access electronic data.
-
Californians approved Proposition 24, which expands the power of the state’s Consumer Privacy Act. The initiative creates a government agency to enforce state privacy protection laws and gives consumers more control over how companies handle their data.
- Portland, Maine, passed a ban on facial recognition technology use by public officials. Portland joins San Francisco, Oakland, Boston and Cambridge, Massachusetts, in banning the use of the technology by public officials, but it doesn’t go as far as Portland, Oregon, which has banned the use of facial recognition in public places by private companies as well.
-
California voters approved Proposition 22, meaning gig workers will remain independent contractors. Tech companies such as Uber, Lyft and Doordash spent more than $200 million as part of their effort to pass the initiative. Observers say the price tag could be a harbinger of the tech industry’s future legal and electoral strategy.
-
More: California’s Privacy Law, Explained | Machine Bias
Chinese Fintech Company’s $35B IPO Delayed Due to Beijing Pressure: Last week, the Shanghai Stock Exchange announced it would delay the initial public offering of Ant Group, just two days before the Chinese fintech company was set to go public. The decision reportedly came after Beijing regulators summoned company executives and Jack Ma, Ant’s controlling shareholder and the co-founder of Alibaba, to a “supervisory interview” and unexpectedly tightened regulations on micro-lenders like Ant. Reports indicate the regulatory pressure was a reaction to an October 24 speech in which Ma criticized Beijing regulators and the “pawnshop mentality” of China’s (primarily state-owned) banks. Ant’s IPO was set to be the biggest in history, with analysts predicting the company would raise more than $35 billion in a dual listing in Hong Kong and Shanghai. Observers say Beijing’s decision to disrupt the IPO augurs a reassertion of the Communist Party’s power over China’s ascendant tech industry.
EU Restricts Exports of Surveillance Tech: On Monday, European Union lawmakers reached a provisional deal to increase export restrictions on facial recognition technology, spyware and other “dual use” technologies that can be employed for cybersurveillance. The new export controls require companies to obtain licenses before selling certain products abroad and to ensure those products won’t be used to violate human rights. The new requirements also are intended to increase transparency by mandating that EU member states either disclose the details of their cyber-surveillance exports or publicly announce their decision to withhold details. That provision is meant to make reporting inaccuracies, which have reportedly plagued previous export control attempts, less likely. The new restrictions will take effect once they have been formally endorsed by the International Trade Committee, the European Parliament and the Council of the European Union.
Using AI to Solve Important Equations 1000 Times Faster: Last month, researchers at Caltech proposed a deep-learning method for solving partial differential equations 1000 times faster than traditional approaches do. PDEs are well-suited for describing and forecasting changes in important natural phenomena such as fluid dynamics, air flow, heat transfer and seismic activity, but are notoriously difficult to solve. As computational power (aka compute) has increased, forecasting models that rely on PDEs — such as those that predict the weather — have become much more accurate. However, due to PDEs’ complexity, only supercomputers have the compute needed to deliver the most accurate results. By increasing the speed of PDE solutions with a lower error rate than other deep-learning methods, the Caltech method may further increase the accuracy of forecasting models and put accurate PDE modeling within reach of smaller research programs and firms.
|
|
Government Updates
Anticipating the Biden Administration’s AI Policy: The election of former Vice President Joe Biden as 46th President of the United States may mean changes in federal policy on artificial intelligence. Here’s how the candidate and his campaign discussed the matter during the race:
-
The Biden campaign’s “Made in America” plan called for a $300 billion investment in R&D and breakthrough technologies, including AI.
-
Last year, Biden pledged to work with allies to “draft a new strategic concept for NATO that acknowledges the challenges of ... disruptive technologies like artificial intelligence.”
-
In a written response to the Council on Foreign Relations, Biden said: “The United States should lead in shaping the rules, norms, and institutions that will govern the use of new technologies, like Artificial Intelligence. Through diplomacy and development finance, we can work with democratic allies to provide countries with a digital alternative to China’s dystopian system of surveillance and censorship.”
-
In an essay for Foreign Affairs earlier this year, Biden wrote, “I will make investment in research and development a cornerstone of my presidency, so that the United States is leading the charge in innovation. ... When it comes to technologies of the future, such as 5G and artificial intelligence, other nations are devoting national resources to dominating their development and determining how they are used. The United States needs to do more to ensure that these technologies are used to promote greater democracy and shared prosperity, not to curb freedom and opportunity at home and abroad. ... As new technologies reshape our economy and society, we must ensure that these engines of progress are bound by laws and ethics.”
In September, CSET published AI policy recommendations for the next administration. Read the suite of five policy one-pagers here.
Senate Releases FY21 Draft Appropriations Bills: On Tuesday, the Senate Appropriations Committee released its Fiscal Year 2021 spending bills, several of which propose investments in AI-related activities. The FY2021 Commerce, Justice and Science Appropriations bill directs $8.5 billion to the National Science Foundation, a nearly $200 million increase. It allocates $6.9 billion of this funding to “maintain NSF’s core basic research portfolio and [support] the priorities of quantum computing and artificial intelligence research.” The Defense bill, meanwhile, includes $104.1 billion in base funding for research, development, test and evaluation accounts, a decrease of $900 million from last year. The committee also recommended full funding for operational systems development at the Pentagon’s Joint Artificial Intelligence Center. House and Senate leaders are working to reach agreement before current federal government funding expires on December 11.
New Head of JAIC Says Its Priorities Are Shifting: Four days before the above-mentioned Senate spending bills were made public, Lt. Gen. Michael Groen, the JAIC’s new director, said the organization would now emphasize the implementation of AI tools rather than focus primarily on developing them. Groen made the comments during an event held by the Center for Strategic and International Studies. The shift — which Groen called “JAIC 2.0” — comes two years after the center’s founding. The new direction of JAIC is meant to help coordinate the Pentagon’s diffuse AI efforts, which Groen said had been “very uneven.” It does not appear that JAIC is abandoning development and acquisition entirely, however; he made the comments just a week after JAIC issued plans to expedite funding for the purchase of commercial AI technology for DOD use.
DARPA Aims to Use AI to Identify Influence Campaigns: The Defense Advanced Research Projects Agency issued a Broad Agency Announcement calling for proposals to help “detect, characterize, and track geopolitical influence campaigns with quantified confidence … using automated influence detection.” The announcement says current methods, which are largely manual, are insufficient to deal with the volume of communications and “lack explanatory and predictive power for deeper issues of geopolitical influence.” DARPA expects to grant multiple awards for the project, “INfluence Campaign Awareness and Sensemaking” (INCAS), around July of next year.
|
|
In Translation
CSET's translations of significant foreign language documents on AI
Xi Jinping Speech: Certain Major Issues for Our National Medium- to Long-Term Economic and Social Development Strategy. This speech, given by Chinese President Xi Jinping in April 2020 but not published until November, lays out perhaps the clearest picture thus far of Xi's proposed "dual circulation" economic strategy. On the one hand, Xi advocates stimulating Chinese consumer spending and accelerating import substitution to reduce the Chinese economy's dependence on foreign trade. On the other, he urges strengthening other countries' dependence on Chinese technology so that China can threaten to cut off their supply when necessary, as a form of deterrence. Paradoxically, Xi also criticizes other countries for "politicizing" or "weaponizing" supply chains in the context of the COVID-19 pandemic.
|
|
What’s New at CSET
REPORTS
PUBLICATIONS, PODCASTS AND VIDEOS
FORECASTS
EVENTS
- On October 29, CSET and Syracuse’s Institute for Security Policy and Law hosted a virtual symposium, National Security Law and the Coming AI Revolution. The discussions featured James Baker, Jason Matheny, Reginald Brothers, Tarun Chhabra and Margarita Konaev.
- On November 10, CSET hosted the most recent webinar in our series on security and emerging technology, Understanding U.S. Military Investments in AI, with Margarita Konaev and Reginald Brothers, in which Konaev discussed her new series of briefs, U.S. Military Investments in Autonomy and AI: Costs, Benefits, and Strategic Effects.
IN THE NEWS
|
|
Upcoming Events
-
November 2-13: United States Strategic Command, 2020 United States Strategic Command Deterrence Symposium featuring Ryan Fedasiuk
-
November 11: National Science and Technology Medals Foundation, Innovation Unscripted: Artificial Intelligence
-
November 12: CSIS, Doubling Down on China, Inc.: An Initial Analysis of China’s 14th Five-Year Plan
-
November 16: UC San Diego, School of Global Policy and Strategy, Report Launch and Press Conference — Meeting the China Challenge: A New American Strategy for Technology Competition featuring Jason Matheny
-
November 16-18: Georgetown University, Nuclear Security Summit featuring Dewey Murdick and Reginald Brothers
-
November 17: Information Technology & Innovation Foundation, Manufacturing Workforce Development and Strengthening Manufacturing Supply Chains: What Can States Do?
-
November 20-22: Internet Law & Policy Foundry, ILFP Policy Hackathon
-
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.
|
|
|
|
