An AI advisory body established by the U.S. Congress will examine how the technology can be applied on the battlefield.
The National Security Commission on Artificial Intelligence was established by lawmakers in the most recent version of the Defense Department’s budget authorization bill. The Pentagon spending bill earmarked $10 million to fund the commission, which will assess how the U.S. military can harness emerging AI and machine learning technologies.
Members of the 15-member AI panel appointed by the Senate and House Armed Services committees, DoD and the Commerce Department, includes senior executives from leading U.S. technology companies, including panel chairman Eric Schmidt, former Google CEO. Robert Work, former deputy secretary of defense, serves as vice-chairman.
Work also heads an AI task force established last year by the Center for Strategic and International Studies. The panel released a report concluding that “machines will be capable of handling more sophisticated tasks in more complex environments, sometimes aiding human decision-making and sometimes operating autonomously.”
DoD is reportedly seeking to shift as much as $70 million in funding to establish a joint AI center.
The congressional action responds to rapid advances in AI technology as well as an ambitious Chinese national AI strategy that seeks to catch and surpass the United States by 2025. While Beijing is pouring billions into AI research, some observers note that China’s AI goals are mostly “aspirational.”
Besides Schmidt and Work, members of the new AI panel include: Eric Horvitz, director of Microsoft Research Labs; Amazon Web Services CEO Andrew Jassy; Andrew Moore, who heads Google Cloud AI; and Oracle CEO Safra Catz.
The Defense Advanced Research Projects Agency and the National Science Foundation moved last year to create a senior-level panel to help coordinate government and industry AI research. DARPA has since announced a $2 billion initiative aimed at developing the “next wave” of AI technologies.
Among the focus areas is so-called “Machine Common Sense,” an elusive category falling somewhere between “narrow” and “general” AI. The approach seeks to train models “to perceive, understand, and judge things that are shared by nearly all people….” The area is seen as ripe for investigation since current frameworks are widely considered “brittle” and lacking in semantic understanding.