In Mid-September 2023, a group of six academic resear- chers from Harvard Business School, The Wharton School, Warwick Business School, MIT Sloan School of Management, and three management consultants of the Boston Consulting Group published what has since becomethethirdmost-downloadedandquotedscholar- ly paper of 2023. “Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality,” short, the “experiment,” as some of the authors call it, is a first-of-its-kind randomized control trial with more than 750 BCG consultants worldwide as subjects.1 It is the first study to test the use of generative AI in a professional services setting—through tasks that reflect what know- ledge workers do every day. “This is important because understanding the implications of LLMs for the work of organizations and individuals has taken on urgency among scholars, workers, companies, and even govern- ments,” the authors explain.2
They were correct in that assumption: After only a few weeks, “Navigating the Jagged Technological Fron- tier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality” has pro- foundly impacted, e.g., the U.K. government’s thinking and decision-making.3 Its conclusions have reached the “AI Safety Summit” that hosted 28 governments and nu- merous industry and civil society experts recently at Bletchley Park. The study, led by Karim Lakhani of Har- vard Business School, has been discussed by C‑suite exe-
- 1 Dell‘Acqua, Fabrizio und McFowland, Edward und Mollick, Ethan R. und Lifshitz-Assaf, Hila und Kellogg, Katherine und Rajendran, Saran und Krayer, Lisa und Candelon, François und Lakhani, Karim R., Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality (15. September 2023). Harvard Business School Technology & Operations Mgt. Unit Working Paper No. 24–013, see SSRN: https://ssrn.com/abstract=4573321 or http:// dx.doi.org/10.2139/ssrn.4573321.
- 2 Dell‘Acqua, Fabrizio and McFowland, Edward and Mollick, Ethan R. and Lifshitz-Assaf, Hila and Kellogg, Katherine and Rajendran, Saran and Krayer, Lisa and Candelon, François and Lakhani, Karim R., Navigating the Jagged Technological Frontier: Field
cutives worldwide and quoted numerous times in newspapers.4
When has a German or European scholarly research paper on AI last had this real-world impact? What is more, the report by Lakhani et al. is only the latest ex- ample of such impactful work with solid influence on companies and governments: the newest thinking co- ming out of the Belfer Center at Harvard Kennedy School on biosecurity on the age of AI by Janet Egan and Eric Rosenbach, published in early November 2023, is set to structure the debate on biological weapons and AI5. Similarly, the Yale Information Society Project at Yale Law School has been owning the discussion on free speech and social media for years now. Especially when it comes to digital policy and digital government, AI po- licy and regulation, and bio- and cybersecurity, U.S. aca- demic institutions have long coined a very different style of research and teaching that has made them global thought leaders and, in fact, agenda-setters for govern- ments and companies, on these digital topics. Even when it comes to such core European topics, like regulating AI, e.g. with the European AI Act, American voices coin the debate almost more than European voices: The letter de- manding a moratorium on AI research for six months and strict regulation, signed by 30,000 experts, resear- chers, industry figures and other leaders in March 2023, among them Danielle Allen, Elon Musk, Geoffrey Hin- ton, and many other prominent voices, was published by the Future of Life Institute in California, led by Anthony Aguirre, the Faggin Presidential Professor for the Phy- sics of Information at U.C. Santa Cruz.
Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality (September 15, 2023). Harvard Business School Technology & Operations Mgt. Unit Working Paper No. 24–0131, page 2.
3 https://www.gov.uk/government/publications/frontier-ai-capabili- ties-and-risks-discussion-paper/future-risks-of-frontier-ai-annex- a.
4 For a short summary of the results, please see https://www.bcg. com/publications/2023/how-people-create-and-destroy-value- with-gen-ai.
5 https://www.belfercenter.org/publication/biosecurity-age-ai- whats-risk.
Why U.S. Universities have more influence in the global debate on AI Governance and Regulation and how German Universities can reclaim their seat at the table. A workshop report
Ordnung der Wissenschaft 2024, ISSN 2197–9197
2 ORDNUNG DER WISSENSCHAFT 1 (2024), 1–6
While this short “Werkstattbericht” or workshop re- port does not presume to explore every angle of the dif- ferences between U.S. and German, or more broadly, Eu- ropean academic institutions, when it comes to teaching and researching digital and technology policy, it never- theless wants to shed light on some of the reasons why we so often find U.S. academic voices at the helm of the- se topics, steering the discussion, and not seldomly stee- ring governments or international bodies like the Euro- pean Union and United Nations. Let’s give some concre- te examples.
II. Not learning for school but for life
To begin with, three characteristics of the collaboration between Karim Lakhani and others with Boston Consul- ting Group make it a prime example to illustrate the enormous differences between U.S. academic institu- tions and German universities and academic institutions when it comes to researching and teaching the societal and policymaking implications of Artificial Intelligence, in particular Generative Artificial Intelligence, or short GenAI.
First, “Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality” was con- ceptualized first and foremost with practical applications and recommendations for corporates, policymakers, and only then other academic researchers in mind. Lakhani and his colleagues primarily answer an exceedingly time- ly and relevant question for corporates, namely whether or not it is worthwhile, from a cost-benefit perspective, to invest in a much-hyped but expensive and complex, potentially even dangerous technology, and if so, how to do it effectively. It is a question that is asked daily by C- suite executives worldwide: what implications does Ge- nAI have for my strategic workforce planning?
Second, Boston Consulting Group, a strategy consul- ting firm that advises C‑suite executives, found not only the perfect study object as a global company of 30,000 + employees with varying backgrounds, seniority levels, and abilities but also an ideal multiplier for the results. The same experiment in a purely academic setting done with university students would not have had the same impact or significance, as the authors acknowledge themselves: “A crucial feature of our experiment was the availability of our experimental subjects. Specifically, we
6 Dell‘Acqua, Fabrizio and McFowland, Edward and Mollick, Ethan R. and Lifshitz-Assaf, Hila and Kellogg, Katherine and Rajendran, Saran and Krayer, Lisa and Candelon, François and Lakhani, Karim R., Navigating the Jagged Technological Frontier: Field
tapped into a high human capital population, with parti- cipants who were not only highly skilled but also enga- ged in tasks that closely mirrored part of their professio- nal activities“.6
Furthermore, the experiment by Lakhani et al. deli- berately highlights starting points to help policymakers gauge where they need to focus policy programs, which are supposed to help those negatively affected by the technology. The paper first gives fact-based and practical insights into who these people may be that require help and who the stakeholders may be that need to be brought to the table to tackle the problem: “An immediate danger emerging from these findings, for instance, is that peop- le will stop delegating work inside the frontier to junior workers, creating long-term training deficits. Navigating the frontier requires expertise, which must be built through formal education, on-the-job training, and em- ployee-driven upskilling.”
Only as an afterthought do the authors want to con- tribute to a purely academic debate. But their first and foremost ambition is to shape the discussion in industry and governments.
These characteristics of “Navigating the Jagged Tech- nological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality,” namely aiming for immediate practical applica- tion of the research in companies, picking strong busi- ness partners and leveraging them not just for research but also for marketing, and, lastly, very clearly stating the broader utility for governments, very much highlight and demonstrate the typical approach of the U.S. profes- sional school. We might add a fourth one: working with practitioners, regardless of their academic references. While, for example, BCG has its research unit and inter- nal think tank with the Bruce Henderson Institute, this is not an academic institution, nor does it claim or want to be. Yet its leaders, seasoned practitioners of AI and Ge- nAI implementation in corporates, are equal co-authors of the scholarly paper — nothing you often see in German academic circles.
III. Characteristics of the U.S. Professional School
All of these characteristics are typical for U.S. professio- nal schools. These schools, like Harvard Business School, but also its more policy-oriented sibling Harvard Kenne- dy School, or, a bit further south of the U.S. East Coast,
Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality (September 15, 2023). Harvard Business School Technology & Operations Mgt. Unit Working Paper No. 24–013, p.17.
Rulf · U.S. Universities influence on AI Governance and Regulation 3
Yale Law School, or, to venture more to the U.S. West Coast, the Goldman School of Public Policy at the Uni- versity of California Berkeley, are not well understood in Germany at all. There is not even an excellent German translation for the “Public Policy” discipline – it is cer- tainly not “Politikwissenschaft.”
The most significant differences – and secrets of suc- cess for why they have so much of a seat at the table in global discussions – to German universities of these U.S. professional schools can be found in their syllabi, in their teaching personnel, and (mainly as a consequence of the latter) their attitudes towards collaboration with private sector actors and governments.
Take, for example, the syllabus of Harvard Kennedy School, arguably the most famous and respected public policy school in the U.S., which has been the alma mater of presidents of the likes of Barack Obama, Ellen John- son Sirleaf, Felipe Calderón, let alone dozens of minis- ters in any country of the world, U.S. congressmen and ‑women, as well as senators, and leaders of the World Bank, IMF, and United Nations. Despite its evident suc- cess, Harvard Kennedy School’s syllabus would hardly get academic approval from a German university presi- dent. I have often experienced a slight haughtiness among German academics when it comes to Harvard Kennedy School classes like “policy analysis,” “leader- ship,” “negotiations,” “and the making of a politician,” and their curricula: efficient, little to no written home- work of academic nature, almost no traditional teacher- centered “chalk-and-talk” teaching, but instead students are put, e.g., through real-time, real-stakes negotiation practices with peers, have to found companies or NGOs, write and pitch op-eds that are published in newspapers around the world, calculate budgets and make trade-offs – in short, student have to put themselves, their visions, and their arguments on the line in real-world situations that prepare them for the careers that they aspire to: dip- lomats, politicians, policymakers, agents of change in ci- vil society organizations. Even lawyers: classical research or time in the library, as German undergraduate or mas- ter students still experience it for the majority of their classes, is not considered appropriate or sufficient to pre- pare for a career as a judge or attorney at internationally renowned institutions like Yale Law School, Harvard Law School or Columbia Law School. Any U.S. law school has at least a law clinic for students to act as legal counsel in real life and practice their skills. Classes are highly interactive and challenging rhetorically; they
7 https://www.hks.harvard.edu/courses/science-and-implications- generative-ai.
mostly center around the latest news and case studies rather than theoretical frameworks.
This practical approach to a profession is particularly relevant regarding a fast-moving topic like digital and technology policy. Consider the fact that any book, even any paper or regulation, like that E.U. AI Act, that was written before November 2022, the release of ChatGPT 3.5 by OpenAI, has almost no relevance anymore for today’s debate on AI, its governance or societal implica- tions. And this is not the first time that technological progress outruns policymakers. In the U.S., governments — federal, state, and local – and universities have learned during the Cold War and its constant nuclear threat that they need to think and debate interdisciplinary if they want their debate to be able to keep up with technologi- cal progress. Furthermore, they need to be current and not recur to frameworks that may no longer be applicable.
Consequently, digital policy is taught differently in these schools than in Germany and Europe.
Firstly, in most U.S. universities, digital and emerging technology policy have their home in the professional school, i.e., in a policy or law school, often have dedica- ted study tracks and are always taught by an interdiscip- linary team and in the case method, i.e., along a practical example of their application. Take, for instance, the new course “The Science and Implications of Generative AI” at Harvard Kennedy School: it is taught by three profes- sors – one economist, one mathematician, and a public policy professor. They promise their students they will learn “through case studies, simulations, and project- based assignments to assess the advantages and risks of deploying generative AI. The curriculum underscores the significance of informed policymaking in this rapid- ly evolving field, seeking to ensure that HKS graduates can harness AI technology responsibly for the benefit of society.”7
By contrast, only some European universities offer interdisciplinary teaching on AI or case methods. Ox- ford University, for example, focuses on the social sci- ence of the internet and digital technology at the Oxford Internet Institute, but through a very academic lens.
ETH Zurich in Switzerland interestingly houses in- terdisciplinary research on the societal implications of new technologies, including AI, in the Department of Humanities, Social and Political Sciences. But in the core European Union itself, despite the E.U. being the first mover on comprehensive legislation on AI with the E.U.
4 ORDNUNG DER WISSENSCHAFT 1 (2024), 1–6
AI Act, only a handful of universities offer interdiscipli- nary classes on AI, among them Technical University of Munich in Germany, the KTH Royal Institute of Techno- logy in Sweden, Delft University of Technology in the Netherlands, and University of Helsinki in Finland. But we have yet to see any of them have as broad and promi- nent a seat at the table as Harvard or Yale have regarding AI policy in Washington. Or a paper that is more broad- ly agenda-setting and globally discourse-dominating than the one from Harvard Business School.
Another huge difference is the formal qualification of teaching personnel and faculty: U.S. professional schools often care more about real-world experience than acade- mic accolades. This goes for all disciplines, really:
Jacinda Ardern, former prime minister of New Zeal- and, is equally part of the Harvard faculty as was Ban Ki- moon, Secretary General of the U.N. Emma Sky, the founding Director of Yale’s International Leadership Center, served as political advisor to the Commanding General of U.S. Forces in Iraq, as development advisor to the Commander of NATO’s International Security Assis- tance Force in Afghanistan, and as political advisor to the U.S. Security Coordinator for the Middle East Peace Process. None have a Ph.D. or would qualify for a formal teaching position in Germany. Similarly, the current ad- ministrator for USAID, the U.S. Agency for International Development, and former United States Ambassador to the United Nations, Samantha Power, who is on leave from not one but two professorships, the Anna Lindh Professor of the Practice of Global Leadership and Pub- lic Policy at Harvard Kennedy School and the William D. Zabel ’61 Professor of Practice in Human Rights at Har- vard Law School, was a practicing journalist before she became one of the most popular professors at Harvard. She also has no Ph.D. degree in sport, let alone a habilitation.
In digital and emerging technology policy, picking the best person for the job today allows U.S. professional schools to attract the most seasoned practitioners as teachers, who bring their experience directly from the front and often still practice while teaching classes. In addition, they can also quickly and fast adapt to new topics.
Bruce Schneier, for example, likely the globally most renowned cybersecurity expert, who is a daily consultant to governments around the world, does not have a doc- toral degree, which would probably take him out of the running for a faculty position in any German university
or academic institution. But it makes him a highly sought-after teacher at Harvard who always contributes the latest insights to his students and decision-makers in Washington.
Similarly, Nick Sinai joined Harvard in 2014 from the White House, where he was the U.S. Deputy Chief Tech- nology Officer. Sinai led President Obama’s Open Data Initiatives, co-led the Open Government Initiative, and helped start the Presidential Innovation Fellow program. Before this, he played a key role in crafting the National Broadband Plan at the FCC. Today, he works as a senior advisor at a Venture Capital firm. However, he still teaches every spring at Harvard Kennedy School a high- ly practical class called “Tech and Innovation in Govern- ment.”. Students there are paired with governments and public sector entities to solve real-world digital prob- lems, like coding a database and designing a digital government solution.8
Consequently, these professional Schools have consi- derable advantages in contributing meaningful research and educating tomorrow’s leaders who already have real- life experience coming out of university. A significant benefit, especially regarding fast-moving topics like Ge- nerative AI, is for both students and professors and com- panies and societies. At the same time, research by pro- fessors is, in turn, inspired by problems from the real world. The study by Lakhani et al. is the latest, but by far not the only example, of them setting the agenda for governments or companies.
This brings us to the last and likely most controversi- al difference between U.S. professional schools and their digital policy work compared to German or European programs: the highly contested topic of industry collabo- ration and sponsoring.
Stanford, Carnegie Mellon, MIT, Harvard and Yale have a long history of collaborating with big tech compa- nies and corporations. Vice versa, Alphabet, the parent company of Google, collaborates with various universi- ties globally through its subsidiaries like Google and DeepMind on AI research and projects. Meta, Microsoft, Amazon – all the large tech companies have university partnerships in the U.S. and their research labs. These collaborations might involve joint research projects, aca- demic grants, fellowship programs, and other forms of scholarly engagement to advance the state of the art in AI and promote the responsible use and understanding of AI technology. OpenAI, still partially a non-profit orga- nization, often collaborates with researchers from diffe-
Rulf · U.S. Universities influence on AI Governance and Regulation 5
rent institutions and may form partnerships with univer- sities for particular projects or initiatives.
They accept money from big tech or other industry collaborations of different forms, e.g., with companies like Boston Consulting Group. It still sometimes raises eyebrows in the German academic community and with good reason. Debates around the economic implications of AI regulation, including its impact on innovation, competition, and market dynamics and discussions of AI’s impact on labor markets and how law can address potential job displacement may be feasible in an ivory- tower setting. But questions around privacy and data protection, e.g., analyzing the sufficiency of existing pri- vacy laws and how they apply to AI, and debating whe- ther new privacy frameworks are needed, or issues of se- curity and cybersecurity of LLMs, e.g., the unique secu- rity challenges posed by AI, and how regulation can mi- tigate risks such as adversarial attacks, or, indeed, a proper assessment of technical standards, e.g. the role of technical standards in AI regulation, and how academic research can contribute to the development of robust, widely-accepted standards – these topics cannot be dis- cussed without collaboration with the developer compa- nies themselves.
While it is unlikely that we will see German academic institutions turn into full-on professional schools, besi- des the few existing initiatives like the Hertie School in Berlin, the Willy-Brandt-School, or the Bucerius Law School in Hamburg, and while we can even argue whe- ther or not that might be sensible on the whole, I stron- gly believe that German and European academic research needs more of a seat at the table, when it comes to tech- nology and digital policy and global debates around regulating technologies like Artificial Intelligence. And that this will only come about by opening up more to the practical, to practitioners as teachers, and to industry as collaborative partners. Besides, it means becoming faster in publishing well-founded statements in more accessib- le publications and giving in to more marketing, also through industry partners.
The example of the U.S. professional schools and their approach shows that these organizations often en- gage with policymakers, academics, technologists, and
the public to foster a better understanding of technology’s impact on individuals and communities and advocate for policies that ensure technology serves the broader public interest. They are crucial in informing and sha- ping the discourse around technology and society in the USA. Through their various programs and initiatives, they each seek to bridge the gap between academic re- search and policy practice and to foster a well-informed public discourse on critical global issues.
And that, after all, is what we need in Germany and Europe, too, when it comes to critical technologies like Artificial Intelligence. Furthermore, we need the next ge- neration of academics to be better trained to bring their arguments into the public domain. With technology like GenAI that has so much potential to cause democratic destabilization and disinformation, it needs trusted voices that know how to communicate clearly and give practical advice to industry, society, and governments.
Kirsten Rulf is a core member of the Technology & Digital Advantage practice at the Boston Consulting Group, as well as a leader on the Financial Institutions team for BCG X, the firm’s tech build and design busi- ness. At BCG, Kirsten focuses on the safe and respon- sible development and implementation of AI and generative AI business models at scale. Her primary fields of expertise are global AI regulation and gover- nance; data governance; the geopolitics of tech; and crafting and implementing data-driven business models. In addition to her work at the firm, Kirsten teaches AI governance and digital transformation at Yale University and is a UC Berkeley Tech Policy Fel- low.
Prior to joining BCG, Kirsten was senior digital policy advisor to German Chancellors Angela Merkel and Olaf Scholz and the Head of the Digital and Data Department at the Federal Chancellery of Germany for more than four years. In that role, she co-negotia- ted the EU AI Act, Data Act, and all other European digital regulation, and was responsible for Germany‘s strategic positioning and global investments in digi- tal technology and infrastructure.
Before her work at the Federal Chancellery, Kirsten taught AI and compliance at Harvard Law School and ran a research group on autonomous vehicles at Har- vard Kennedy School. Before that, she was a TV corre- spondent for the BBC and for German national TV ARD and its flagship news bulletin Tagesschau.
6 ORDNUNG DER WISSENSCHAFT 1 (2024), 1–6