Vitalik Buterin Discusses AI 2040 Debate, Open Source, and Potential Slowdown Triggers

Vitalik Buterin said in a post on X that AI 2040 and its critics hold incompatible worldviews about the pace and importance of AI progress. According to Odaily, he described AI 2040 as arguing that some form of superintelligence will emerge across many scenarios by 2040 unless strong measures fully stop it, while critics argue AI 2040 underestimates human coordination and threatens freedom but do not treat artificial superintelligence itself as a risk of power concentration.

Buterin said his position would shift depending on what he believed: if current AI is merely ordinary technology, he would be closer to the critics; if superintelligence is likely to arrive by 2030 by default, he would be closer to the AI 2040 camp. He added that because uncertainty remains high, he is staying open to slowing down or pausing AI development.

He also said he is uncomfortable with a stance he attributed to some large AI companies and intellectuals that “open source is unfavorable” and that an ideal outcome would be for their side to hold global controlling dominance.

Buterin said one reason he supports the d/acc platform is that several directions are worth advancing under either worldview, including formal verification, cryptography, secure and open hardware, pandemic resilience, defensive biotechnology, food and basic resource security, public epistemic systems, and non-power-concentrating physical security.

He added that the 2040 plan has become more supportive of open source and has incorporated an idea he described as “mutually assured compute destruction,” which he said is an improvement over allowing a small number of actors to selectively deprive targets of rights.

On whether to slow down or pause, Buterin said there is no way to avoid trade-offs, but it is possible to set trigger conditions in advance and remain more open to slowing or pausing if sufficient conditions are met within a specified time window.

He also said that if he were Elon Musk or Mark Zuckerberg, he would more extensively reshape Twitter into a platform that helps identify and facilitate large win-win agreements and bring more people into the discussion, though he said this may be naive. He said he currently does not see any non-naive plan for managing a transition to artificial superintelligence and therefore tends to show some tolerance toward those attempting to address it.