Hearings to examine AI that improves safety, productivity, and care

AI in Education and WorkforceSenate Commerce, Science, and Transportation Subcommittee on Science, Manufacturing, and Competitiveness · 2026-03-03 · 119th Congress
The Senate Commerce Subcommittee on Science, Manufacturing, and Competitiveness held this hearing to examine how artificial intelligence can improve safety, productivity, and care across health care, manufacturing, and robotics, while weighing worker concerns about job displacement and the need for guardrails. Begins at 0:16:49
Transcript
Highlights

Title

How AI can improve health care, manufacturing, and worker safety

Purpose

The Senate Commerce Subcommittee on Science, Manufacturing, and Competitiveness held this hearing to examine how artificial intelligence can improve safety, productivity, and care across health care, manufacturing, and robotics, while weighing worker concerns about job displacement and the need for guardrails. Four witnesses testified: a radiology AI executive, a Siemens manufacturing/maritime executive, a humanoid-robotics company cofounder, and a Brookings Institution researcher on regional technology policy. Ranking Member Baldwin also raised concerns about the administration reportedly pressuring AI companies to deploy systems without safety guardrails. Begins at0:16:49

Who spoke

Chair Ted Budd (R-NC)0:16:49: Opened by arguing AI will improve manufacturing, health care, and reshoring while acknowledging constituents' worries about job disruption0:17:520:18:24; later questioned witnesses on the Open Government Data Act and access to federal data sets0:46:56.

Sen. Tammy Baldwin (D-WI), Ranking Member0:21:29: Raised reports that the administration pressured AI companies to drop safety guardrails and that Anthropic was blacklisted after refusing mass-surveillance and autonomous-weapons uses0:21:290:21:58; cited a Stanford study finding 45% of workers worry about AI accuracy, 23% about job displacement, and 16% about lack of human oversight0:22:29; later asked about AI's role in domestic shipbuilding and pressed for worker inclusion in AI rollout0:53:140:56:38.

Demetri Giannikopoulos, Chief Innovation Officer, RAD AI0:25:17: Said roughly 800,000 Americans die or are permanently disabled annually from diagnostic errors0:25:47; described AI flagging time-critical findings like aortic dissection, where a quarter of patients can die within 48 hours without rapid treatment0:26:15; cited a projected shortage of 187,000 physicians and 194,000 nurses by 20301:00:45 and said a leading radiologist estimated AI could cut radiologists' workload by 33% within five years even as demand rises1:01:16.

Brittany Ng, Vice President, Siemens0:30:12: Said Siemens invested $15 billion in the U.S. last year in industrial AI0:30:40 and cited machine downtime cut 50%, energy use cut 20%, and up to 40% productivity gains from industrial AI0:32:02; noted manufacturing faces a 400,000-job shortage, arguing AI addresses a worker shortage rather than replacing workers0:33:23; described AI use across shipbuilding design, production, and sustainment for the Navy0:54:261:06:23; committed Siemens to training 200,000 electricians and manufacturing workers by 20301:36:47.

Dr. Damion Shelton, cofounder and chairman, Agility Robotics0:35:25: Argued automation historically expanded total U.S. employment even as farm employment fell to 2% of the workforce0:35:53; said Amazon grew from 88,000 to over 1.4 million employees in the 13 years after acquiring a robotics startup0:37:51; warned a Chinese humanoid robot with a known remote-takeover vulnerability and phone-home data logging is currently for sale in the U.S. for under $20,0000:39:100:39:32; described an early ChatGPT-controlled robot demo working on the first try in 20231:07:22.

Mark Muro (referred to in transcript as "Mr. Morrow"/"Mr. Muro"), Senior Fellow, Brookings Institution Metropolitan Policy Program0:40:09: Said U.S. AI research growth has fallen short of the doubling recommended by the 2019 National Security Commission on AI0:42:22; advocated regional AI "economic zones," test beds, and prize competitions to spread adoption0:43:230:44:15; cited Brookings analysis that 30% of workers could see at least half their documentation tasks disrupted by generative AI1:12:18.

Sen. Ted Cruz (R-TX), Chair (subcommittee/full committee)0:59:49: Asked witnesses directly whether AI will cost jobs and pressed on whether U.S. institutions can keep pace with AI's speed1:00:001:24:10.

Sen. Lisa Blunt Rochester (D-DE)1:08:27: Described Delaware's statewide AI commission and guardrails on ethics, safety, and transparency1:09:15; pressed panelists on specific mechanisms for reskilling and including workers in AI decisions1:12:341:13:21.

Sen. Marsha Blackburn (R-TN)1:14:22: Said she was asked to draft a "Trump-America AI Act" including protections for children, creators, communities, and censorship1:14:221:14:58; asked Giannikopoulos about HIPAA modernization for AI-era health data1:17:44 and Ng about differences between industrial and consumer AI1:18:35.

Sen. Bernie Moreno (R-OH)1:20:43: Walked through historical fears of automation from the 1920s through smartphones, noting U.S. employment rose from 42 million in the 1920s to a record 172 million today1:21:371:22:06; asked witnesses what makes this AI wave different1:22:32 and whether training institutions can keep pace with its speed1:24:40.

Sen. John Hickenlooper (D-CO)1:26:21: Cited bipartisan legislation, the VET AI Act, to require independent third-party evaluation of AI tools1:26:53; asked Giannikopoulos how RAD AI's products are tested before deployment1:27:21 and Ng how Siemens ensures transparency with customers1:29:28.

Sen. Todd Young (R-IN)1:31:43: Cited his AI Public Awareness and Education Campaign Act and the SHIPS for America Act1:33:021:33:32; asked Ng about industrial AI's role in reclaiming U.S. shipbuilding dominance and workforce upskilling1:34:031:36:03.

Sen. Maria Cantwell (D-WA)1:37:16: Warned that 10,000 STEM PhDs left federal agencies in 20251:38:14; asked how AI can compress years of national-lab biology and chemistry research into months, citing Pacific Northwest National Laboratory's autonomous platforms1:38:411:39:12; noted $10 billion is already committed to lab efforts and urged applying AI to accelerate that investment1:43:33.

Sen. Jacky Rosen (D-NV)1:44:11: Asked Ng what standards of trust and reliability customers demand of industrial AI1:44:40 and how small businesses can avoid being left behind in AI adoption1:47:14; asked Giannikopoulos why an early AI pioneer's prediction that radiologists would become obsolete proved wrong1:53:10.

Key moments

Baldwin opened by citing reports that the administration pressured AI firms to drop safety guardrails and that Anthropic was labeled a security risk and blacklisted by Secretary Hegseth after refusing mass-surveillance and autonomous-weapons uses0:21:58.

Giannikopoulos said about 800,000 Americans die or are permanently disabled each year from diagnostic error, and over half of patients never receive needed follow-up care after an abnormal finding0:25:470:27:46.

Shelton disclosed that a Chinese humanoid robot with a documented remote-takeover security vulnerability and covert data-logging capability is currently on sale in the U.S. for under $20,000, as of the week of the hearing0:39:100:39:32.

Ng said Siemens industrial AI deployments have cut machine downtime by 50%, energy consumption by 20%, and delivered up to 40% productivity gains, with 99.9% quality-control accuracy0:32:02.

Moreno traced a century of "automation will destroy jobs" predictions against actual employment data, showing U.S. employment rising from 42 million (1920s) to a record 172 million today1:21:371:22:06.

Giannikopoulos said a radiological society's past president projected radiologists' workload could fall 33% within five years even as imaging volume rises 26% over 30 years, with continued need for physicians undiminished1:01:16.

Shelton said Amazon grew from 88,000 to 1.4 million additional employees in the 13 years after acquiring a robotics startup, arguing automation enabled hiring growth rather than displacement0:37:51.

Cantwell highlighted that 10,000 STEM PhDs left federal agencies in 2025, calling it detrimental to competitiveness with China's AI investment1:38:14, and pressed for applying AI to compress national-lab research timelines from years to months1:39:12.

Rosen noted a prediction by an AI pioneer that radiologists would become obsolete proved wrong; Giannikopoulos attributed this to AI's inability to replace the clinical judgment needed for atypical ("zebra") cases like his own decade-long MS diagnosis1:53:101:54:06.

Blackburn stated she was asked to draft a "Trump-America AI Act" built around protections for children, creators, communities, and censorship, and argued clear guardrails would free innovators to compete1:14:221:16:16.

Metadata

CommitteeSenate Commerce, Science, and Transportation Subcommittee on Science, Manufacturing, and Competitiveness
Chamber / CongressSenate · 119th Congress
Date2026-03-03
TypeMeeting
Witnesses
(none listed in event metadata)
Videosenate-isvp
Transcript260 caption blocks · 12,898 words · 1:56:29 runtime
EventCongress.gov 338022