Hurd, Kelly: Government must pay more attention to AI
- By Matt Leonard
- Sep 25, 2018
A new congressional report on artificial intelligence calls on the government to increase its spending on AI research and development and focus on bringing more AI skills into the federal workforce.
The report, released on Sept. 25 by the House Oversight and Government Reform Subcommittee on Information Technology, makes broad recommendations that could be used to guide legislation in future Congresses, subcommittee chairman Rep. Will Hurd (R-Texas) said in a call with ranking member Rep. Robin Kelly (D-Ill.) and reporters on Tuesday.
“It’s a race,” Hurd said. “We all know this and one of the things we need is a national strategy.”
What exactly that national strategy will look like has yet to be determined, but Hurd said it could follow in the footsteps of strategies used for quantum computing or nanotechnology.
The report is the result of a series of hearings held by the subcommittee on AI, how the government could benefit from it and its potential impact on the workforce.
The report suggests working off of the recommendations provided in previous AI studies released under the Obama administration, adding that the “current Administration should prioritize building on these efforts and ensure federal agencies are implementing these recommendations.”
It also said the current administration has taken important steps to make AI a priority, such as creating the National Science and Technology Council Select Committee on AI.
The Obama-era reports released included numerous recommendations, many of which overlap with what’s outlined in this new report from the subcommittee. Those commonalities include increased research funding, a need to better understand potential workforce impacts, a review of privacy implications and the creation of clear data standards.
This latest report recommends the federal government bring in more workers with AI skills, but it doesn’t outline any steps to make this happen. An importance should be placed on education, Hurd said, adding that AI skills should be attainable at an undergraduate level and not require a PhD.
As for research, Hurd said understanding the potential for bias within these technologies is an important funding priority. “Understanding bias is critical for the future of this technology,” he said.
Data management and storage are other areas that need more research, Kelly said.
During the subcommittee's first AI hearing this past February, Ian Buck, the vice president and general manager of accelerated computing at NVIDIA, suggested the government could help advance AI by opening up more of its datasets.
“Data is the fuel that drives the AI engine,” Buck said. “The federal government has access to vast sources of information.”
The new report comes to the same conclusion, suggesting that “thousands of different data sets” could be opened up to the public for the benefit of AI researchers. To make this happen, the subcommittee urged the Senate to pass the OPEN Government Data Act, which the House passed last year. It is unlikely, however, that the Senate will pass that bill before the end of the year, Hurd said.
A second hearing in March focused on how the government could benefit from these technologies. Tools like robotic processes automation, which automate digital tasks, could prove especially beneficial, according to the report.
“The government is already using AI,” Kelly said. “We just need to focus on increasing effectiveness and efficiency.”
And while legislation to sharpen that focus is unlikely this year, Hurd said the report was written with a longer view. “This is a topic that is going to … be important beyond this Congress," he said. "[W]e’re trying to set a foundation with the work we’ve done … for other committees to run with.”
Matt Leonard is a former reporter for GCN.