Intro
The centre for future generations did some important thinking on how a CERN for AI could function. Like, in practical, mechanistic detail. We should thank them.1
As always, somebody dislikes that others can just do stuff. In this case:
Ah, the myth of academic credentials being a necessary hallmark of a good idea. It is all too common. Not just in Australia — Professor Toby Walsh F-A-A F-T-S-E’s home — but especially also in Europe.
Outsmarting ourselves without disempowering all of humanity will require immense effort. Superhuman AI would likely be our last invention. A lot of genius is required for a handover of the economy to AIs.2 Let’s maybe dedicate all available brain power to advancing safely.3
We need to build hard and think even harder - from hardware to software security, information security, and societal resilience. Humans, even our smartest, build and think better when heavily clustered. Europe has untapped potential for harboring such an international endeavour. Unfortunately, right now, Europe actively scatters most of its smartest people (or worse, provides reasons to leave).4
Time to dispell some myths, lest we’ll all have to move to Washington DC or the San Francisco Bay Area soon. You can still just do stuff in Europe if you look closely enough. The fact that many people aren’t doing stuff is but a cultural failure, compounded by overupdates on WWII. Stop being scared of yourself and start building the society that holds the balance.
Despite current policy failures, Europe is closer to a “CERN for AI” than Professor Toby Walsh F-A-A F-T-S-E from Australia can imagine.5 I’d like to help make it happen and thought CERN’s conception is a good starting point. But most of the official explainers weren’t adding up — something diplomatic was going on. I figured I could provide some better sketches due to our work experience.
So, let's look at how CERN got done, ackshually.
Big CERN
Here’s my summary of how CERN explains its history in an impressive act of historical diplomacy:
Raoul Dautry, Pierre Auger, Lew Kowarski, Edoardo Amaldi, and Niels Bohr spent the 1940s dreaming of Making European Science Great Again.
At the 1949 European Cultural Conference in Lausanne, 57-year-old French Nobel laureate Louis de Broglie proposes the creation of a “multi-national laboratory”.
Five months later Isidor Rabi, another Nobel-prized American colleague, authorizes UNESCO to initiate a regional research lab.
21 months later, 11 countries sign an agreement birthing the acronym ‘CERN’.
Three months later, they start writing a convention and scouting grounds — Danish, Dutch, Swiss, and French — for the location, quickly settling on Geneva.
Another 13 months later and the financing and a convention are hashed out — it’s time to build (and ratify)!
From a speech in December 1949 to the complete ratification of the convention, it took almost 4.5 years for the twelve founding states to negotiate and ratify the now-called “European Organization for Nuclear Research”.6
Imagine everything that must have happened in 26 months of acronym-pregnancy. And then another 16 months of labor(ious discussion). Oh the poor surrogate! So uh, what actually took so much time? It must have been hard to figure out whether everyone wants the child in the first place. And the conception doesn’t exactly sound immaculate… Let’s look behind the thin veneer of geopolitical finesse.
Epistemic status: ~propaganda. Instead of post-rationalizing a research approach, indulge me for storytime. I’m putting all of the relevant public sources into this footnote.7 Systematic referencing would have kept me from publishing indefinitely. It would also mostly have been cargocult science. This piece is a patchwork of inferences made from 20h of conscious research, plus intuitions shaped by our work. Wherever there are links in the text, it’s links that I only use once.8 Imagine every other claim links to one of the sources in footnote 7.
The anecdata
The official 1949 proponent of CERN, Louis de Broglie, was an elder French statesman of physics. In revolutionary French fashion, Louis was concerned with decline. This lesser-known Louis appears to have been concerned with the vanishing grandeur of European physics in particular. And in classical fashion, this Louis also did not care to step down but contented himself with lending prestige from the ivory tower. In short, he didn’t do much for CERN beyond lending his name to the proposal.
As far as I can tell, the heavy lifting for CERN was done by its first and only "General Secretary": Edoardo Amaldi. 16 years De Broglie’s junior, Amaldi put together many conceptual pieces and maintained the critical network. His motivations appear to me as highly underrated for anyone attempting similar feats. They might also explain why even CERN’s Wikipedia only mentions him once: he understood that he was part of something much bigger.
To Amaldi, CERN was about the future of humanity.
At the very start of his career, Amaldi immediately experienced true human excellence as part of Enrico Fermi's "Via Panisperna boys". They discovered slow neutrinos - crucial for sustaining nuclear chain reactions. The basis for the cleanest and safest source of quasi-abundant energy! Imagine the hype of making this discovery - VPboiz4lyfe!
Unfortunately, but to no surprise, Fermi’s wife and two out of seven VP boys were Jewish and the Nazis were just getting started. Between 1936 and 1938, three VP boys leave for North America. Another one “disappears” after four years of depression.9
With the Via Panisperna boys broken up, Amaldi is left yearning for a breakthrough squad. Despite briefly getting drafted to the African front, Amaldi mostly manages to stay out of the war. Rather than sitting it out, he even attempts to exploit fascist megalomania: in the lead-up to the 1942 World's Fair, he pitches a particle accelerator to Mussolini’s government (the fair was canceled in June 1941).
Amaldi views science as the ultimate force for cross-cultural convergence and seeks to cluster his colleagues. In 1942, his idealism makes him decide to avoid fission research to stay out of war efforts. Despite dwindling opportunities, Amaldi feels a strong duty to give back to his motherland and to shepherd the survival of peaceful, Italian science.
In 1946, Fermi offers him a professorship in Chicago. Shocked by Fermi’s secrecy, due to purported military interests, Amaldi rejects yet again. For the next years, Amaldi and his colleague Gilberto Bernardini try to rebuild the Italian physics community. On a tight budget, this means doing things at small scale and many of the most talented colleagues continue to leave for the much better resourced US.
To stop the brain drain, a monumental project would be needed. Witnessing US nuclear research on a trip in 1946, the idea of a European accelerator comes up again. Once back back from his trip, Amaldi’s research group examines the infrastructure needs and costs of building and running a facility. Amaldi is convinced that real science has to happen without military involvement.
While focused on building up Italian physics, Amaldi - now the de facto face of Italian post-war physics - continuously advocates for a shared accelerator. But scientists around Europe lack the political insight or ambition to pull it off. It takes France - harboring ambitions to become a nuclear power - to publish a national investigation into the military facilities necessary to build out energy and weapons.10
In December 1949 at the European Cultural Conference in Lausanne, a French government official reads out a statement from Louis de Broglie that purportedly “initiates” CERN. That’s right, de Broglie, claimed to be one of the main characters of CERN’s official history, wasn’t seen anywhere near CERN’s founding.11
A month before de Broglie’s announcement, various Nordic countries, led by Niels Bohr in Copenhagen, had advanced a similar proposal for collaboration. Everyone agreed that American expertise is needed. But most thought the Americans aren’t willing to help.12 Neither the Nordics, nor the French, planned to include Western Germany.
Meanwhile, an old friend of Amaldi — Bernardini — moves to the US and shares Amaldi’s accelerator plans with Isidor Isaac Rabi, who had built up the Brookhaven National Laboratory and worked in Los Alamos. Well aware of geopolitical realities, Rabi sees Amaldi’s plans as an opportunity to stabilize Europe in service of the US.
To ensure the Euros include West Germany and slow down European nuclear weaponization, the US acts fast. The communists are feared to gain ground in French politics. To rein in Europe’s potential, the Marshall Plan has to succeed at reintegration. And as the Soviets are now clearly on track for the bomb, the fear of handing out nuclear secrets is also diminished.
A facility the scale of CERN — without test reactors — would slow down the spread of nuclear weapons while accelerating science that accelerates American progress, too.
Cognizant of French political force, Rabi engages with fellow physicist Pierre Auger, director of UNESCO's science division close to French politicians. With Auger’s institutional backing, Rabi convinces the US Foreign Service to sponsor a UNESCO resolution he drafted - plausibly with help from the Department of Defense.
Before presenting the UNESCO resolution, Rabi meets with Amaldi. Somebody neutral and competent has to spearhead the implementation. Amaldi is a qualified, reliable candidate with a vision, technical understanding, and a large network.
At the fifth UNESCO General Conference, held in Florence in June 1950, Rabi tables a resolution for UNESCO to "assist and encourage the formation of regional research laboratories…"
The next years, Amaldi and Auger campaign through Europe. On their grand tour, they convince governments and academics alike that the ambitions laid out are worth pursuing. Niels Bohr, biased by his personal plans, questions the feasibility of applied research at such scale. The British question the opportunity cost. But Amaldi’s plan and — occasionally passion — eventually convince the British nobility.
Having secured key political allies, buy-in eventually cascades in 1952. After plenty of work sessions and council meetings, it’s official: the council moves from Copenhagen to Switzerland in 1954. The construction begins. CERN’s history, from here on, is much less abstruse.
Shortly after launching CERN, Amaldi — of course — gets involved in Pugwash. He’s a genuinely hardworking good guy, this Amaldi.
So what to make of this?
I.e. what do I make of this?
CERN was an outlier in diplomatic success - and accurately attributing credit can be opposed to getting things done and maintaining their function.
US defense interests are powerful, yes. But they’re also insufficient. Even right after WWII, among the allies, it wasn’t obvious how to put 2+2 together.
The French played a key role in CERN’s conception only in so far as their push for nuclear weapons likely made it easier for Rabi to convince DoD of support. In the actual conception of CERN, however, the French only seem to have played a secondary role.
The US did well not to figure prominently in CERN’s inception but leverage proxies. It was probably important for the US to stay out of the annals for the French to assume full ownership. This might successfully have delayed French nuclear weapons by 2-10 years.
While Amaldi was in some sense “exploited” by the US, it is easy to underestimate his role in pushing things through. Anecdata says he was the convincing force, not Auger. I guess that fighting for due credit is simply not fun and would have kept him from doing more useful things.
The cosmopolitan won because he didn’t let pride get in his way. He didn’t bother asking for respect. He simply did it because he thought it was right.
But let’s get back to where we started: academic credentialism and building CERN 2.0.
The above is evidence of how excellence has little to do with journals or conferences. Beyond a modicum of intelligence, doing great things mostly has to do with doing the hard work, getting through to the right people, and showing integrity. Including in politics. In the olden days, few people were privileged. Nowadays, however, you don’t even need a university degree to obtain these features and go do stuff.13
Listening to the experts certainly isn’t always the best path either: CERN’s leadership currently hopes to build a >$20bn toy for… their comfy retirement. You’d think all that compute could be put to better use. Switzerland might also find a better way to remain internationally relevant. Let’s not focus any of Europe’s smartest people on what looks likely to be quite literally a pipe dream.
It is the next Amaldis that will make CERN for AI happen. The De Broglies and credentials are still helpful, of course, but in a world of digitized trust networks, they’re becoming less and less crucial.
Amaldi was able to take an immense amount of responsibility on himself because he had community support and a clear vision: the securitization of science in the US would not help bring about a better world. Whether he was right is a topic for another time.
At CERN's official 30th anniversary in 1984, Rabi said:
I expected that Europe, which was the cradle of science, once brought back into the path, would achieve some very great things. I hope that the scientists at CERN will remember that they have other duties than exploring further into particle physics. They represent the combination of centuries and centuries of investigation study, and scholarship to show the power of the human spirit. So, I appeal to them not to consider themselves as technicians but as guardians of this flame of European unity so that Europe can help preserve the peace of the world.
Thank you, cfg! Now we can have more productive discussions about which aspects concretely need workshopping.
All available brain power that’s not already used for crucial societal functions, of course. Even if you include the private sector, “core AI people” alone won’t cut it. We need all hands on deck for this transition.
Yes, the French are an exception. No worries, enough of this post is about the French, we’ll get there.
With all due respect: Toby, maybe it’s time to move. Glorifying last century's academia is pretty “non-core AI”.
OERN wasn’t as swanky, so they kept CERN - to my big disappointment.
https://cds.cern.ch/record/151765/files/CERN-CHS-1.pdf
https://cds.cern.ch/record/151766/files/CERN-CHS-2.pdf
https://cds.cern.ch/record/151764/files/CERN-CHS-3.pdf
https://sci-hub.se/https://doi.org/10.1007/s00016-004-0225-z
https://pubs.aip.org/physicstoday/article/58/4/87/394407/Cern-s-Early-History-Revisited
https://cerncourier.com/a/edoardo-amaldi-a-true-statesman-of-science/
https://ep-news.web.cern.ch/content/interview-ugo-amaldi-reflections-cerns-70-year-legacy-and-european-collaboration
As you might have figured out by now, other footnotes and links are mostly tongue-in-cheek or dangerous rabbit holes. You’re welcome.
According to my questionable research, yes, this is a euphemism for “one of the most promising physicists of the last century jumped off a ferry to kill himself”.
Yes, eerily similar to the AI sitch. But beware the man of one story.
Yes, the homie Louis wasn’t even at the conference! The French used his name for their strategic advantage.
DéjAI vu much?
Source: it was revealed to me in a dream. Maybe in the 1940s, mostly academics, aristocrats, clerics and a few industrialists were privileged enough to have time to think about mundane things like “do I think better when surrounded by all of my smartest friends?” With a few decades of additional progress, these deep insights even came to high school dropouts in Australia. If someone could please get it published at NeurIPS, Toby Walsh might appreciate learning about it, too.
Great post and welcome on Substack! I guess one interesting bit to keep in mind is US export controls. For original CERN, having the Germans and the Italians in meant having reactors out. Does it matter that some European countries are Tier 2 in Diffusion Bill? (I genuinely dk)