Donald Trump’s first term as president coincided with the beginning of three important trends that continue to shape the AI world: the rise of transformer-based models, surging AI investment, and a more hawkish foreign policy approach to China. But important questions remain about what Trump’s second term will mean for AI. In this excerpt from policy.ai — CSET’s newsletter on artificial intelligence, emerging technology, and security policy — Alex Friedland explores what Donald Trump’s presidency will mean for AI policy, semiconductors, and private AI investment.
What a Second Trump Term Means for AI Policy
Donald Trump’s victory in this month’s presidential election has many in the AI world wondering: what will the next four years look like for AI policy? Trump’s first term as president coincided with the start of the current boom in AI research and development. In 2017 — his first year in office — Google researchers published “Attention Is All You Need,” the hugely influential research paper that introduced the transformer architecture underlying today’s most powerful generative AI systems. In 2020, near the end of his first term, OpenAI released GPT-3, the company’s first model to make serious mainstream waves (a later version of GPT-3 powered ChatGPT when it launched in 2022).
The Trump administration didn’t ignore the burgeoning technology: In 2019, President Trump signed the country’s first executive order on AI, which laid out a strategy to promote U.S. leadership in AI development. Then in December 2020, he issued the first executive order on federal government AI use.
Just as the intervening four years have seen an unprecedented surge in AI development and investment, there has been a corresponding boom in AI policy efforts from the White House under Joe Biden. Highlights include:
- In 2023, President Biden issued an executive order that, among other things, imposed reporting obligations on developers of the most powerful AI systems, set off a flurry of AI-related actions among federal agencies (see our tracker for more), and set up the National Artificial Intelligence Research Resource pilot program.
- The National Institute of Standards and Technology released its voluntary AI Risk Management Framework to help AI developers manage risk (see our writeup from last year) and the Office of Science and Technology Policy published its non-binding “Blueprint for an AI Bill of Rights” (see our writeup from 2022).
- Among a host of AI-related actions, the Pentagon: stood down its original AI-focused organization — the Joint Artificial Intelligence Center, originally established in 2018 under President Trump — and consolidated its data, AI, and digital efforts under a new Chief Digital and Artificial Intelligence Office; launched the “Replicator Initiative” to field thousands of low-cost, attritable, autonomous systems on an accelerated timeline (Replicator 2 was announced earlier this year); and updated its autonomous weapons policy.
- Late last month, the White House issued a National Security Memorandum on AI to guide the U.S. national security strategy toward AI. The memo lays out a comprehensive plan to maintain U.S. leadership in AI while ensuring its safe and responsible use across national security agencies (read reactions from CSET experts).
On the campaign trail, Donald Trump said he would repeal Biden’s AI executive order on “day one” of his term, a goal that made its way into the Republican Party’s official 2024 platform. As Mozilla Fellow Deb Raji pointed out on social media, the Biden administration’s executive branch-heavy approach to AI policy — while more easily enacted than a traditional legislative strategy — is much more easily undone than policies that have passed through Congress.
With Republicans’ victory in both the House and Senate, it seems unlikely that much of Biden’s AI agenda will be resuscitated on Capitol Hill. While both chambers of Congress have pursued some notable bipartisan AI initiatives in recent years — including Senate Majority Leader Chuck Schumer’s AI Insight Forums and House Speaker Mike Johnson and Democratic Leader Hakeem Jeffries’ bipartisan AI Task Force — neither chamber has shown much appetite for sweeping legislation.
What a Second Trump Term Means for Semiconductors
Since taking office in 2021, President Biden’s administration has pursued a two-pronged semiconductor strategy: incentivizing domestic manufacturing through the CHIPS and Science Act and limiting China’s access to high-end chips with strict export controls.
On export controls, it seems unlikely that President Trump will roll back the Biden administration’s restrictions — announced in October 2022 — that aimed to significantly cut off China from the semiconductors needed for high-end AI applications. Those controls were part of a trend that began under President Trump’s first administration to restrict China’s access to critical technology, and Trump’s picks for key positions in his new administration — including noted “China hawks” Marco Rubio and Michael Waltz — don’t indicate a change in tack.
Subsidizing domestic semiconductor manufacturing could be a different story. Though the 2022 CHIPS and Science Act passed through Congress with significant bipartisan support, Trump criticized the law during an October interview with the podcaster Joe Rogan. Soon after, House Speaker Mike Johnson indicated that repealing the law could be on his agenda, but quickly walked back those comments.
While most observers don’t expect the Trump administration to try to repeal the law, the White House is working to allocate and finalize the remainder of the $39 billion in CHIPS Act incentives before President Biden leaves office. Last week, the Commerce Department finalized a $6.6 billion award for TSMC’s investment in new Arizona facilities, finalized $1.5 billion for GlobalFoundries on Wednesday, and is expected to finalize more awards soon.
What a Second Trump Term Means for AI Development
One area where questions remain most pronounced is what Donald Trump’s victory means for AI investment and development. Last year, the United States led the way in private investment — of the $95 billion invested worldwide, the United States accounted for $67 billion of it.
From one angle, it looks like a second Trump term could be a boon for the technology companies financing the AI boom: President Trump earned the support of some of the tech world’s most outspoken voices, including Tesla, SpaceX, and xAI CEO Elon Musk and prominent venture capitalists like Marc Andreessen and David Sacks. After the election, Andreessen — who has championed a “Techno-Optimist” agenda and pointed to concerns about “hostile” overregulation as a key driver of his political activities — said Trump’s victory “felt like a boot off the throat.”
But it’s far from clear that Trump’s second term will be a bacchanal for AI developers. During the first Trump administration, the Justice Department and Federal Trade Commission launched antitrust cases against Google and Facebook, respectively, and Amazon alleged it lost out on a lucrative DOD cloud computing contract due to political differences between the president and Amazon founder and then-CEO Jeff Bezos. Vice President-elect JD Vance has expressed support for current FTC Chair and Big Tech critic Lina Khan, and Attorney General nominee Matt Gaetz has previously called to break up Big Tech companies.
Time will tell which voices — the tech accelerationists or the tech skeptics — will win out in the second Trump administration.
This newsletter excerpt is from the November 21, 2024, edition of policy.ai — CSET’s newsletter on artificial intelligence, emerging technology, and security policy, written by Alex Friedland. Other stories from this edition include:
- AI Wall? Hints of Diminishing Returns, But Developers Don’t Seem Worried
- Big Tech’s LLMs for the Pentagon (and the People’s Liberation Army?)
- Anthropic Paces the AI Agent Race with Computer-Using Claude
Read the full newsletter and subscribe to receive every edition of policy.ai.