Ethereum co-founder Vitalik Buterin has once again outlined his principled stance on the development of artificial intelligence. This time, his criticism is directed at the largest AI companies, whose ambitions, in his view, border on outright dreams of "world domination." At the center of the discussion is the "AI 2040" scenario, which Buterin believes represents a choice between a "naive" and a "naive squared" approach to managing the transition to artificial superintelligence (ASI).
According to my analysis, the essence of the dispute boils down to a fundamental divide in the perception of AI's pace and risks. The "AI 2040" scenario assumes that ASI will become a reality by 2040 under almost any circumstances, unless humanity makes titanic efforts to completely stop it. Critics of this scenario, Buterin observes, accuse it of a naive belief in people's ability to coordinate. But the developer himself counters: they are naive to think that the transition to ASI will go smoothly and without conflict by default.
Buterin emphasizes that the critics' position only makes sense in a world where AI remains an "ordinary technology." However, in a world where ASI is possible as early as 2030, this criticism simply does not hold up. Vitalik himself admits uncertainty: he cannot say with confidence which of these worlds we live in. If he were sure of at least one scenario, he would join the corresponding camp. This admission is a key signal for the market: even the leading minds in the industry do not have a unified vision of the future.
Concerns about power concentration
Buterin is particularly concerned about the rhetoric of giants like OpenAI and others. He is alarmed by the stance that "open source is bad, and a good outcome is when our guys get controlling world domination." In his view, in a "normal" world, such statements should trigger all political alarm bells at once. He sees superintelligence itself as a colossal risk of power concentration.
That is why Buterin consistently promotes the d/acc (defensive acceleration) platform. This concept includes the development of formal verification, cryptography, secure open hardware, pandemic resilience, and public epistemology. The value of these areas is that they are useful in both future scenarios—whether AI develops quickly or slowly. Notably, Vitalik positively assessed that the "AI 2040" plan has become more friendly to open source and even mandates its use.
Seeking a compromise
Buterin sees the key element of his strategy as finding a deal acceptable to both sides. He proposes agreeing in advance on a set of triggers signaling that "something serious is happening." As examples, he cites superpandemics, an unemployment rate above 25%, or the emergence of autonomous weapons.
If a sufficient number of such triggers are activated within an agreed timeframe, the parties pre-agree to seriously consider slowing down or pausing development. The idea is that critics will accept this, expecting the triggers not to fire, while the concerned will accept it, expecting the opposite. Buterin even addressed the heads of tech giants, stating that if he were Elon Musk or Mark Zuckerberg, he would restructure social platforms to seek such mutually beneficial deals. However, he himself admits that this idea is also naive.
My analysis: Buterin's position is essentially a call to decentralize the management of ASI risks. In a world where cryptocurrencies and blockchain offer alternative governance models, his criticism of AI giants is not just a philosophical debate but a defense of the basic principles of distributed power. The market should closely monitor the development of d/acc, as these technologies could become the bridge between the crypto world and the safe development of AI.