House AI report makes recommendations supporting R&D, workforce, and AI small business integration

The federal government spending on non-defense AI R&D has increased from $560 million in fiscal year 2018 to $2.1 billion in 2023, according to the bipartisan House Task Force on Artificial Intelligence report released in the closing month of the 118th session. The report encourages Congress to continue federal R&D efforts, with specific recommendations for Congress to support fundamental R&D for continued global leadership in AI innovation, increase technology transfer from university R&D to market, and promote public-private partnerships for AI R&D. The report also encourages Congress to align national AI strategy with broader U.S. technology strategy, explore how to accelerate scientific discovery across disciplines with AI, and support AI R&D for small businesses.

Findings and recommendations are distributed across 15 topics, including government use, privacy, national security, civil liberties, intellectual property, authenticity, R&D, energy requirements and AI application within small business, agriculture, health care and finance.

Opportunities for expansion of government efforts that complement state and regional TBED initiatives abound, if implemented. For instance, the report notes that AI R&D would be bolstered by providing university researchers greater access to advanced computational resources. The task force members write, “AI research at universities continues to be limited by access to data and computational power. Even the most well-resourced academic institutions do not have the resources to train AI systems of comparable complexity as the most advanced AI models. … As a result, many researchers have left academia entirely for industry.”

Noting workforce issues, the report authors state that “Currently, the most common pathway into the AI workforce is through a four-year degree. Interest in AI-related degrees like computer science is surging—enrollment in computer science degrees grew 249% between 2011 and 2020.” Despite this growth, the task force sees a growing gap nationally in the talent available and to be required to address research, development and deployment of AI Recommendations outlined for overcoming this gap include facilitating public-private partnerships to bolster the AI workforce, developing regional expertise when supporting government-university-industry partnerships, broadening pathways to the AI workforce, evaluating existing workforce development programs, and supporting NSF curricula development.

The report also considers the challenges for small businesses in adopting AI technology, noting, "While large companies have entire departments dedicated to maneuvering federal, state, and local challenges, smaller companies lose out on potential innovation and must reallocate time and resources to maintain compliance. Discussions about AI regulation are advancing quickly at the federal, state, and local levels, and worries about compliance burden are pervasive among small businesses.” The report urges Congress to address those challenges by providing resources for small business AI adoption and easing compliance burdens.

The task force noted other concerns effecting regional economic opportunity from AI; foremost, perhaps, is that AI’s energy requirements are likely to strain the existing electrical capacity as the development and deployment of AI grows. The report states, “Accompanying the predictions of soaring demand are warnings that the electric grid cannot reliably meet future needs.”

Water demands for cooling in data centers are also a concern mentioned in the report, which states “The average data center has a water usage effectiveness of 1.8 L of water consumed (not recirculated) per 1 kWh of IT equipment energy.” Recirculation can considerably reduce the need for fresh water, but data centers measure their energy use in much larger units than kilowatt hours.

Carbon emissions may be the greatest long-term challenge of the rapid race to be first in AI. Already coming off the hottest year on record, the task force found emissions for some of the largest users of AI and data centers are headed in the wrong direction to reduce the growing negative impacts of the current upward trend in global atmospheric temperatures. “According to their environmental reports, the carbon emissions of technology companies are soaring—Google reports their emissions have increased 50% since 2019, Microsoft reports a 30% increase since 2020, and Meta reports a 66% increase since 2021. And this is projected to continue increasing: these hyperscale companies have already announced plans to build several additional gigawatt-scale (up to 5 GW) data centers to meet the AI demand.”

Many pundits and political analysts focused closely on the federal government anticipate AI will be a priority for the new 119th Congress and returning Trump Administration.