Hallo,
In January of 2026, I contacted, amongst others, the Humboldt Institute for Internet and society (HIIG). This institute, located in central Berlin, aims to improve perspective on digitalization across multiple domains.
A few years ago, as a 17-year-old, I remember having some vague idea that work in a ‘think tank’ is something I might want to do. I did not really know what was involved in ‘working in a think tank’, nor do I remember exactly how I came upon the idea. Nonetheless, when applying for the Laidlaw scholarship in the first place, I remember thinking that this seemed to be a good way into that kind of thing. And so it was.
I sent this cold email specifically to Dr Theresa Züger, whose Impact AI team were working on an assessment framework for ‘AI-for-good’ projects. AI was and continues to be something on many people’s minds. The preceding summer, I had written an essay for a competition about AI companionship. I felt, writing this essay, the muddle surrounding responsible AI use. Millions of vulnerable people appeared to have improved mental health as a result of a false (perhaps delusional) belief in an AI friend[1]. Certain individuals were profiting from this morally convoluted arrangement. What struck me quite forcefully was that the guiding hand in this situation was not the prescriptions of mental health professionals, policymakers, or even undergraduate philosophy students. The guiding hand here belonged to the people who had created the technology. They did not need to listen to the opinions of concerned experts on the periphery. They simply continued in bringing about the world which they wanted, in line with their principles and/or desires. An opportunity to work with a team assessing ‘AI-for-good’ projects felt like it addressed relevantly similar concerns to these. It asked: how responsibly are these people acting? Admittedly I also felt that the fact of having written this essay would be somewhat of a leg to stand on for the cold email, beyond the ‘ox.ac.uk’ email address.
After a thoroughly busy term, and a thoroughly busy two weeks post-term, fellow Laidlaw scholar Albee and I made our way to Berlin. On this journey we perused the literature created and sent to us by the Impact AI team, for preliminary study. Much of this literature focused on the foundation of their method – a functional conception of the ‘good’ constituting the standard for assessment of AI projects. As well as this, the literature revealed in some detail the shape of their method: interviews and surveys with members of AI initiatives, sorted and analysed according to a set of criteria in some way determined by the project’s aforementioned definition of ‘good’.
At this point in time, my conception of the project was resultantly much more focused on this groundwork, as opposed to any practical application - I thought of Impact-AI more as a theory than as a tool.
This conception shifted within the first week of work, when we were assigned AI use-cases to carry out the method on. I will not go into detail on the work I was doing, as I am unsure of how much I am at liberty to reveal. My work mainly consisted, though, in unearthing and framing the associated issues with this project for democracy and governance, and the legwork of coding and analysing the interview data. Before any of this, though, we were asked to produce feedback for some of their unpublished work. Whilst this was largely to ensure that those new to the project could understand there work, our feedback was, in some cases, implemented – particularly that which helped with the interpretability of the graphic, as well as the classic pedantic grammatical and stylistic points. In the main work, I was able to contribute a fully coded set of interviews, which enabled the team to go on writing the report. Alongside this, I produced a document on the question of how my specific AI-initiative in focus related to the public interest, a contentious question which I am still ultimately unsure of.
Through this work, I felt the same questions cropping up again. Who has agency in the development of this technology? What is the distinction between irresponsible and responsible development and use? What power do the people who are interested in responsible development and use have? And so on and so on.
In my hours at the office, I spent a good deal of time thinking about specific people relevant to this question – mainly Silicon Valley founders. This was a common conversation topic in the office. Many had been travelling down this road for a good few years, offering insightful opinions and pieces of information. Conversation also extended to the manosphere, the future of work, shifting international dynamics. A conversation could be struck through just the question – ‘what are you working on, then?’. In hindsight, I do wish that I had asked this question more, though. It was occasionally all too easy for Albee and I to just stick to one another at lunch, if we were feeling less social than normal. I do think we did a good job on the whole, though. Recommendations for Berlin nightlife were also discussed in (sometimes shocking) detail, alongside comparisons between German and UK culture and living standards, as well as comparisons between Berlin and peoples’ home cities. I would say that these conversations within the office outside of work were one of the most valuable aspects of these 6 weeks – one of the many clichés that is a cliché for the good reason that it is so often true.
There were in these interactions a few points of awkwardness. Surprisingly, language was not one of these, although I did feel a constant guilt for my lack of German (‘mein Deutsch ist nicht so gut…’). The most obvious point of slight awkwardness was mine and Albee’s juniority. Quite often Albee and I would be, by default, asked whether we were master’s students or PhD candidates. Quite often, in introduction rounds, fellows and researchers would introduce themselves through highly specific career-long research interests. However, admission of undergraduate-level generality was of course not, as I had on some level feared, met with any kind of belittlement or ‘what-are-you-doing-here?’-ness. Instead, this discrepancy in experience was met with a good deal of grace, and I felt that the people of HIIG often went out of their way to enrich us conversationally as a result.
A highlight of the six weeks was the creation of a multi-media blog series, in partnership with Albee. It was a nice opportunity to digest the intense subject matter of the working week in a non-serious, creative way. I had not had light fun making things for a good few years, directing my creative efforts mostly towards quite rigidly analytic essays for my degree. We certainly took liberties when creating this content. Admittedly, we did generally stray from the prescribed analysis of our time in terms of the character framework. I do feel as if the authenticity of the blog benefited as a result of this decision, allowing us to practise honesty, a key virtue of courage.
Genuinely, though, the whole experience did present lots of insights and issues concerning the nature of leadership. Within the team, Dr Theresa Züger’s presence was certainly noticed when she returned from some brief time off. I felt much more drive in the team when she was present, which I would put down to her infectious commitment to the project. In her time off, other members of the team such as Lena Winter and Dr Laura State were responsible for Albee and I’s work. I was incredibly grateful for how much trust they placed in us to work independently, whilst also making us feel comfortable in asking them for help, when it was often needed. Having been part of the Impact AI team is something that will undoubtedly shape my conduct as a future team member.
More broadly, leadership emerged as a theme in the work itself. I was particularly struck by the case of Palantir co-founder and CEO Alex Karp. I had naively felt that incredibly influential people who I regarded as problematic, even dangerous, would be alien to me. Alex Karp, however, subverted that expectation. There is, amongst many, a general distrust of this company. Beyond their substantive contribution to what many regard as government mandated human catastrophes, there is something cartoonishly evil in their branding. Their adverts glorify a heavily militarised future, the name of the company is quite literally borrowed from the evil spy technology from Lord of the Rings, and so on. Alex Karp is a key proponent of the Technocratic/Western-Supremacist/Mentius Moldbug/Peter Thiel/etc. movement which has gained much influence in the US and beyond. And yet, by the sounds of it, Alex Karp was moving in many of the same circles that would describe Palantir in similarly unfavourable terms to me. His liberal upbringing culminated in a PhD at Frankfurt University, the birthplace of critical theory. During his law degree at Stanford, he would often combat his friend, Peter Thiel’s, conservatism – that’s according to both founders. The point I’m trying to make is that this is a person who has the educational origin story not a million miles from a Laidlaw Scholar, yet many of the hallmarks of a cartoon-evil leader (certainly, he leans into them for his brand). I disagree with his worldview. However, I get the sense that he has considered many if not all of the points that I would put to him. What reasons would he give for, by all appearances, leaving behind the set of ideas with which I currently associate? Is it, in spite of what he may say, personal reasons, like greed or insecurity? Or has his decision to influence the world in this way the product of careful reasoning? These questions are important to me, as they point to a potential fork in the Laidlaw-adjacent road of ethical leadership. I want to be able to say that Alex Karp has veered off course. (Maybe my issue is less about leadership and more about character, or politics. He is a leader, but what I’m interested in is his character.) To clarify, I’m not suggesting that I foresee myself starting a Palantir-esque company. My point is just that my understanding of leadership/character has been shaken by the existence of Alex Karp – that is, I think I need to think much more about what kind of leader/adult I want to become, as it appeared he had similar ideas at one point, but has since changed drastically, and from where I stand this change makes no sense.
Enough rambling! I have left HIIG with more questions than answers. I have certainly enjoyed working in a think tank but am now more curious to see what else there is out there. I am more uncertain about the future of AI, and the future more broadly, as I feel more aware (though still largely ignorant) of the many, many forces influencing this future. I have lots of questions about what its like to live and work in other cities, as well as how think tanks are regarded within influential organisations such as governments and large companies. Also please watch the blogs if you haven’t already as I think they are far more informative than this.
Hasta luego, Laidlaw
Ben.
[1] ‘False/delusional’ supposing that (1) consciousness on both sides of the relation is a condition of 'friendship' and (2) LLMs are decidedly not conscious. This is of course lacking detail, but the point I think is clear.