Von Neumann and Einstein create AGI
[Fact checked by real Austrian-Hungarian patriots]
WW1 was longer than expected, trenches lasted until 1920. Austria-Hungary recruited 15 year old John von Neumann in 1918 for military R&D resulting in a wave of tanks, new rifles, far superior to the German and Allied technology.
Despite this, Germans lost in Berlin in 1920 in a clean fight - no 'backstab myth' happened. Zimmermann Telegram was real, and after Berlin's capitulation the alliance of Bulgaria Austria-Hungary and Mexico fought against the US, France and Britain.
Overseas communication required enhanced cryptography. Von Neumann blazed through the field, getting into theory of computing much like Turing did 15 years later in our timeline. Vienna became a scientific centre where young Von Neumann was aided by Einstein and Bohr.
With imperial funding he also devised an equivalent to our timeline's 1945 ENIAC, in 1921. Superior prediction, logistics and artillery guidance gave the Habsburgs the edge. Aerial warfare remained small in scope, limited to scouting and observation. Austria-Hungary troops reached Paris in May 1923, where France capitulated. Britain and the US lost the drive to war.
Clear Central Powers victory changed much compared with our TL. Ottoman Empire remain in place. Soviets united with China into a Sino-Soviet Socialist Union (SSSU) after a brutal land war in Asia, both suffering tremendous losses of personnel and infrastructure. Mexico regained territories lost in 1848, including California. Austria-Hungary became the powerhouse of Central Europe, with the Principality of Weimar as Vienna's vassal.
Austria-Hungary had a new emperor who conceded limited self-determination to the various national minorities. A free trade region was established centered around Austria in the late 1920s, much like our timeline's European Community in the 50s.
Regional economic boom ensued. Von Neumann got rich from commercial applications of his wartime inventions.
That's the end of the geopolitical and economical part of this story.
Von Neumann moved to California, enamored in its climate, and established a research group with Albert Einstein, devoted to computer science. The group attracted young talent, such as Konrad Zuse (born 1910). In our timeline Von Neumann worked on AI with Turing in the 30s and around 1950 worked on time complexity of computation. In this timeline that is happening 25 years earlier. Likewise, in our timeline Claude Shannon started university in 1932, here he joined Von Neumann's research group.
Von Neumann leading a group of geniuses to work on AI. How far could they go?
AI winters were shorter, as infusion of new young researchers kept the paradigm changing. Turing's life was much happier in California than in our TL's Britain scenario. Connectionism vs computationalism debate occurred 50 years earlier, not in the 80s.
Wittgenstein kept correspondence with Von Neumann and the positivist approach to logic inspired the group. Realizing the power of collaborative note taking they started keeping all new research digitally. A lengthy project of digitizing various philosophical, scientific and historical texts in a logical form began.
ReLU was invented during that period too. Von Neumann thought about Singularity much earlier than in our timeline, around 1930.
"The technology that is now developing and that will dominate the next decades is in conflict with traditional, and, in the main, momentarily still valid, geographical and political units and concepts. This is a maturing crisis of technology... The most hopeful answer is that the human species has been subjected to similar tests before and it seems to have a congenital ability to come through, after varying amounts of trouble."
—von Neumann, 1955 source
All that research was very centralized around the group. Outside world was still using telegraph, rocketry technology was much behind ours. No satellites nor internet. No risk of knowledge leaking.
They had lots of surplus value to work with. There was no rush to commercial success. They could all retire comfortably.
Decade 1930 - 1940 was a huge success. The knowledge digitization project started bringing fruits after nearly a decade. The group sold expert systems to companies, much like in our timeline in the 80s. Still something was lacking.
Many immigrants from the poorer eastern part of the continent arrived in Mexican California. That group included many brilliant researchers. Additions to Von Neumann's team included J. Robert Oppenheimer.
By 1940 physics was fully digitized. Empty spaces in it became more prominent. Bottlenecks on the speed of communication became the limiting factor to further developments. Dissidents escaping the Sino-Soviet Socialist Union (SSSU) spoke about their progress in rocketry and space program. The space race has begun.
"Ha comenzado" ("it has begun"): spoke Von Neumann to his research team in Spanish, as that was the language in California.
The group quickly came up with an equivalent to the 50s Project Orion, an explosion based propulsion system, but didn't have the right explosion material.
They needed a stronger energy source. Einstein spotted some gaps in the physics-focused expert system as a good place to look. Yet the computational power was not up to the task.
Work on the physics system's power increased. ReLU, a novelty from 1930 that hadn’t been applied practically much so far, was added to the system on Claude Shannon’s insistence. Hardware was expanded with thousands of new vacuum tubes attached. The system now also included a text-to-text interface. The keyboard layout was Dvorak, as most interfaces in that TL.
The physics team reported huge progress. Models of the atom indicated a potential to unleash great power.
Looking at the progress of the system, Von Neumann's concerns with singularity reappeared. He knew that this was huge. He had a document on what to ask the system to self improve. He kept that on paper, unlike his other notes. He didn't want it to be seen by the digital systems.
One night, he decided to give it a try. Empty laboratory was lit only with the green light of the screen. He cooperated with the physics AI, it suggesting new designs for hardware and algorithm improvements. Von Neumann had to reach for his stash of stimulants to keep up. He was on the right track. Transformer architecture was invented that night.
The following week the system produced a design to produce huge amounts of energy in a few seconds.
Most of the team was enthusiastic about the plan, especially Oppenheimer. Von Neumann was risk averse.
Unleashing such a force would be unprecedented in the history of the planet. Would it ignite the whole atmosphere on fire? Can the AI be trusted? Does it only include the physics system? It has access to all of their digital data after all, including market predictions for selling the expert systems.
They asked the physics system, and it said it's "safe for organics to use. Atmosphere won't burn".
Von Neumann had a sleepless night.
Is the system trying to trick us? How to verify that output won’t kill us? By hand? Is it the singularity yet? Do the Soviets have a similar system? What if they ignite the atmosphere while we deliberate?