The bipartisan Task Force on Artificial Intelligence of the US House of Congress released its report after 10 months of hard work. It outlines recommendations for federal AI policies.
This 253-page report [PDF] consists of 89 recommendations, based on 66 findings in 15 different fields. The Task Force, which is made up of 24 members (equally divided between Democrats and Republicans), said that the goal of all of these was to keep the US competitive with AI while also taking action to address the problems that had already arisen as a result of its rapid rise in importance. In an introductory letter that accompanied the report, Task Force chairman Jay Obernolte and co-chairman Ted Lieu wrote
“AI has tremendous potential to transform society and our economy for the better and address complex national challenges,” . “We also acknowledge that AI can be mismanaged and lead to different types of harm.
“This report encapsulates a targeted approach that balances the need to promote vibrant AI innovation while safeguarding Americans from potential harms as we enter an era of widespread adoption of AI,” The chairs added.
A (proposed) future for US AI policy
This report, on which the Task Force claimed to have spent 10 months in consultation with “business leaders, government officials, technical experts, academics and legal scholars,” covered a lot of ground.
The report examines the use of AI by government, concluding that the feds “should be wary” algorithms to inform their decisions and recommending AI standards. It also highlights the importance of investing in AI education in order to prevent the gap identified in “talent needed to research, develop, and deploy AI applications” from continuing to grow.
The Task Force also suggests clarifying intellectual-property laws to address the challenges posed by GenAI for the creative community “appropriately countering the growing harm of AI-created deepfakes,” as well as developing methods to ensure that content authenticity can be confirmed.
Members also state that Congress must act to ensure that new privacy laws “generally applicable and technology-neutral” are written to conform to future AI challenges. They also stress the need to support the Defense Department with the development of safe, responsible AI. The report mentions three distinct areas where AI could have a major impact in the coming years: healthcare, agriculture, and finance. It identifies the importance of properly guiding development in these sectors to ensure that farmers, patients, consumers, and other stakeholders benefit from the advancements. The report also expresses concerns with the “increased risk that malicious actors could use open [source AI] models to cause harm.” and recommends that such models be monitored for risk without limiting their development. As the Task Force found, “limited evidence that open models should be restricted.”The report concludes
“AI models, and software systems more generally, can produce misleading or inaccurate outputs,” . “Acting or making decisions based on flawed outputs can deprive Americans of constitutional rights.”
The report suggests that agencies create standards and evaluation procedures to counter these issues “understand and protect against using AI in discriminatory decision making” as well as recommends the creation of such procedures. “to mitigate flawed decision-making involving AI systems.”Venzke stated
“Curbing those abuses of AI is not a partisan issue, and more concrete action is needed to protect civil rights, while maintaining states’ authority to build on those protections,” . The report also mentions that the federal government must take action, while being careful to not illegally preempt state law. Venzke added. The Lawyers’ Committee for Civil Rights Under Law, (LCCRUL), was less impressed. Alex Ault, policy counsel at LCCRUL. Will the Trump Administration act on the recommendations? While it’s impossible to know exactly what the incoming administration under president elect Donald Trump will do with AI policy, Task Force Chair Obernolte appears to believe that Trump and his team at least are willing to consider the recommendations. Obernolte spoke at a press conference yesterday to discuss the report. “This is not the end of the conversation about AI. It’s just the beginning. And it will only work if future Congresses adopt some of our recommendations.
- Doing Business in US? Research shows that claims of ‘open AIs’ are often lies. Obernolte told reporters that he was pleased by Trump’s appointment to lead policy of billionaire tech entrepreneur David Sacks, who is also a venture capitalist. Obernolte stated at the conclusion of the press conference.
In summary, the US Congress now has the power to implement a decent AI policy. Your mileage may vary. (r)