Study Tour
The Economic Power of AI: Balancing Regulation and Innovation
"I don’t know."
This was the most frequent response we received from meeting partners during the recent delegation visit to Washington, D.C., and San Francisco, focused on "The Economic Power of AI." As part of our World Order and Globalization Hub, we convened seven experts from Brazil, Tunisia, Mexico, the Netherlands, Thailand, and Turkey, all specializing in public policy or technology, to examine the impacts of artificial intelligence (AI).
This common response did not arise from a lack of knowledge among the people we spoke to; our meeting partners included a variety of experts—policy advisors, academics, and technologists—who are well-versed in AI. Rather, it underscored the uncertainty surrounding the future of AI. Across Capitol Hill, think tanks, and leading technology companies, the prevailing theme was the unpredictability of AI. Will it displace jobs? Likely. Will it create new opportunities? Likely. Will it reshape society? Undoubtedly. However, the timeline and specific nature of these changes remain unclear.
Despite this uncertainty, our meetings yielded valuable insights into the current landscape of AI and the ongoing discourse about its potentials and challenges, which will be outlined below.
The Situation in Washington, D.C.
During the first half of the week, the delegation engaged in a series of meetings in the nation’s capital, primarily focusing on the politics surrounding AI development, with the exception of a few sessions that delved more into the technological aspects of AI. Our meeting partners in Washington, D.C., included think tanks, government officials, and officials on Capitol Hill.
AI in U.S. Politics
Since the change in administration in January, Washington, D.C. has experienced considerable uncertainty. The rapid restructuring of government agencies and shifts in existing policies have made it challenging to maintain a clear overview of developments, including those related to AI.
The delegation sought to understand how perceptions of AI development have evolved in light of the transition into Trump’s second term. Our meeting partners explained that while the Biden administration adopted a cautious approach with the implementation of the executive order on AI to establish guidelines and restrictions for AI development, the Trump administration has aimed to dismantle these measures, promoting less restrictive development. Although there is welcome support for innovation in AI, the removal of restrictions raises significant ethical, privacy, and security concerns. Despite this, most of our meeting partners remain cautiously optimistic that the implications of the new administration’s changes will not be as severe as anticipated.
A common concern seemed to be regarding politicians not taking AI development seriously enough, as their focus shifts elsewhere. While some bills aim to regulate AI and address its misuse—such as the Take It Down Act which “(...) criminalizes the publication of non-consensual intimate imagery (NCII), including AI-generated NCII (or “deepfake revenge pornography”)”[1]—our meeting partners believe Congress must prioritize AI to a greater extent. A major challenge for this is that many politicians do not fully grasp AI and its potential implications. Efforts are underway to bridge this gap through the creation of AI educational courses but as it is developing at record speed it is difficult for even the courses to keep up. Perhaps due to this lack of understanding of AI’s potential implications, there is a prevailing attitude among politicians that everything will ultimately be fine; however, the reality is more complex. If we cannot fully trust platforms like Meta and the future remains ambiguous, complacency is unwarranted.
Furthermore, even if Congress focuses on AI, achieving bipartisan consensus remains difficult. Although both Republicans and Democrats agree that the U.S. should lead in AI development for public benefit and support educational initiatives related to AI and intellectual property debates—despite differing views on specifics like copyright—there are substantial divisions over clean energy usage, investment levels, and regulatory measures. Additionally, disagreements persist regarding federal oversight of AI and the pre-release requirements for AI systems. Most Democrats advocate for data centers to operate on clean energy and for guidelines to protect citizens, whereas Republicans contend that such regulations may hinder development and disadvantage U.S. tech companies.
Nonetheless, both parties express a desire for the EU to relax its AI regulations, which drive regulatory changes globally, considering it a potential leverage point in trade negotiations. Yet, our meeting partners are optimistic that the U.S. will work collaboratively with other nations, rather than monopolizing AI development as in times of uncertainty, cooperation is the most effective approach to advancing AI technologies. Although the new administration has cut down on foreign aid and investments, our meeting partners agree that investing in development and innovation abroad will ultimately also profit the U.S.
AI and Energy
One of the key topics surrounding AI and the development of new data centers is energy use. Currently, around 4% of U.S. energy consumption is dedicated to powering data centers, highlighting the energy demand from both private and public sectors. This surge in energy consumption raises concerns, particularly in relation to climate action goals. Our meetings with energy experts in Washington, D.C. were especially helpful in yielding answers to the delegation’s questions on AI and energy efficiency.
We learned that nuclear energy is being championed as an ideal solution to manage low growth in energy demand, and efforts are underway to revitalize existing nuclear fleets and construct new ones. However, regulatory challenges and generational concerns present substantial obstacles. Additionally, while experts are exploring AI's potential to positively influence climate goals, the reality remains that clean energy sources face significant challenges, particularly the need for 24/7 operational capability for data centers. This issue continues to be a topic of ongoing discussion among stakeholders. A potential solution recommended by our meeting partners would be a balanced approach that combines fossil fuels with clean energy, along with further developments such as the newly implemented cooling technologies that minimize water usage. Major long-term investments in computing infrastructure are crucial, as energy consumption remains a primary barrier to deploying large-scale systems.
Although optimistic projections suggest that AI’s electricity consumption will only account for approximately 6% of U.S. energy by 2030, some estimates propose it could be up to 12%. Our meeting partners emphasized that as initially inefficient technologies tend to improve over time, this is also the hope for energy use in data centers. However, there is a need for advancements in chip efficiency and a stronger focus on enhancing the efficiency of AI software and algorithms.
Furthermore, data centers could be subsidized or restructured to meet the increasing demand, but this requires shifts in how we project and manage future infrastructure needs.
Although AI’s application shows promise, for example for national challenges, such as long-term energy storage, media coverage continues to criticize the energy use of AI, often failing to recognize the historical context and advancements in energy efficiency.
Yet, improvements in energy efficiency for algorithms may have limits, indicating that while AI has potential, its current trajectory might not meet the expectations for energy efficiency. Thus, the search for innovative solutions must continue.
AI in Defense and Security
Another topic at the forefront of the delegation’s mind was the role of AI in defense and security. With the nature of warfare shifting dramatically, the role of AI in the defense sector cannot be disregarded. This has spurred debates on how AI should be integrated into military operations as a national security asset. According to our meeting partners, leveraging AI to enhance efficiency in defense and security is justifiable, but cybersecurity requirements for companies engaging with the Department of Defense (DoD) are essential. In fact, the DoD is already using autonomous weapon systems, albeit under established guidelines that should be constantly updated to ensure it keeps up with new challenges presented by AI.
AI also plays a vital role in countering cyber threats, as the rise in cyber attacks has coincided with a shortage of cybersecurity experts and AI can significantly aid in identifying cyber threats. In this realm, robust public-private information sharing is crucial to effectively communicate incidents and defenses against attacks.
Addressing Ethical Concerns
Although our meetings revealed significant benefits that AI offers across various sectors, including climate initiatives, public health, education, and energy efficiency as well as aids in preventing wildfires, enhancing national security, and addressing challenges like long-term energy storage and judicial process streamlining, the rapid advancement of AI generates critical ethical questions.
One concern is the impact of data centers on local communities, as seen in Mexico, where a facility was built in a region with limited water availability. When asked about this, our meeting partners ensured that there are regulatory frameworks in place that ensure companies justify their location choices but that there is definitely room for improvement in this area.
Another essential point is trustworthiness in AI, with definitions for what that means varying globally. While some amount of ambiguity is good for speeding up development, recognizing that there is ambiguity and understanding who is involved in creating certain systems is crucial. The trustworthiness of AI also opens up questions about whether and to what extent humans should be kept in the loop when it comes to AI. Most of our meeting partners agree that while too much human input on AI will slow down the process of development as the speed and scale of systems make having humans in the loop unattainable, humans do need to be a part of the design and impact assessment to ensure that standards are being upheld and ethics are not violated.
Achieving trustworthy and human-centric AI is challenging, and establishing benchmarks for safety and accountability is critical, particularly for autonomous technologies like self-driving cars or weapon systems.
While our meeting partners in Washington, D.C. support maximizing data utilization, they believe that data protection remains a pressing issue and that it is important for democratic countries to safeguard their citizens’ right to privacy. The need for transparency on what data AI is trained on is something both Republicans and Democrats agree on. To support AI companies that face constraints on data usage while enhancing their operational efficiency, it is vital to explore methods for scaling data that maintain privacy protections.
Some predict that machines will surpass humans in both blue- and white-collar jobs in the near future, raising important ethical concerns about job displacement. Our meetings in Washington, D.C., helped us gain a deeper understanding of the implications that could arise when AI becomes advanced enough to begin replacing roles like those of teachers. Such a shift would not only transform our society but would also have profound effects across various sectors, necessitating solutions for not just displaced employees, but for society as a whole.
Furthermore, some of our meeting partners expressed concerns about AI's potential influence on biology. While there are positive possibilities, such as curing diseases like cancer, there is also the risk of negative applications, including the weaponization of diseases and pandemics.
The delegation’s talks with academics highlighted the concern for education amidst the rise of AI. Apps and online programs such as Data Canvas are taking over higher education and as students’ ability to concentrate and practice critical thinking has noticeably declined, new solutions are needed. The transformation of education shows a need for tech literacy to foster critical thinking among students.
In summary, while AI's potential is vast, its risks cannot be overlooked. We must prioritize responsible AI development that ensures democratic values and public welfare are upheld. A balanced approach is key to maximizing benefits while mitigating risks.
Silicon Valley’s projections for AI
One of the states that has pushed for more AI legislation is California. This is no surprise, as this is also where most of the tech originates. To find out what the tech experts had to say on AI, the delegation travelled out to San Francisco, CA and Silicon Valley for the second half of the week and spoke to founders of AI companies as well as other experts in AI.
Our meeting partners in Silicon Valley generally agree that there is a need for some regulations in the field of AI and that they must be globally aligned, however, they also stress the importance of unrestricted AI research to maintain a competitive edge, particularly in the US-China context. Especially in the Bay Area, which attracts companies with its spirit of innovation, the balancing of innovation with regulation is crucial, as overregulation may hinder advancements.
The delegation had a lot of questions for our meeting partners regarding the future of AI and the next steps in its development. One of our meeting partners shared insights on how AI is increasingly integrated into daily operations across companies, streamlining processes and enhancing efficiency in sectors like marketing. Another meeting partner discussed their work on fourth-generation AI, preparing for quantum computing, and developing Self-Sovereign Identity for secure digital identity management. Additionally, there is a strong focus on enhancing reliability systems. Efforts are underway to address and rectify data quality issues, which will ultimately lead to more reliable AI outputs in the long run.
A significant challenge facing companies today is the need to develop solutions that will remain effective over the next five years. The emphasis is on creating adaptable products that can evolve alongside the changing data landscape and technological advancements, ensuring their continued relevance and value.
Overall, the atmosphere surrounding AI in the Bay Area is optimistic. While concerns about overregulation persist, the state has promised to assess the situation. The spirit of innovation is so robust that regulatory measures may struggle to keep pace with the rapid advancements in the field and that any potential regulations will be moderated.
Conclusion
The implications of AI for our future remain uncertain. While it is clear that AI is evolving rapidly and will transform our world, the extent of this change remains undetermined. Policymakers need to start focusing on AI, and democratic nations should collaborate to find ethical solutions that uphold democratic values. Balancing regulation and innovation is crucial to protect citizens while promoting progress. Despite the U.S. following Trump’s ‘America First’ approach, there is hope for greater global cooperation in AI moving forward.
[1] Senate Commerce Committee. (2025, March). House Leaders Pledge to Advance 'Take It Down' Act at Sen. Cruz's Bipartisan Roundtable with First Lady Melania Trump. https://www.commerce.senate.gov/2025/3/house-leaders-pledge-to-advance-take-it-down-act-at-sen-cruz-s-bipartisan-roundtable-with-first-lady-melania-trump