FULL STORY

'AI 2040' Vision Released Sparks Doom Debate

The release of 'AI 2040: Plan A' sparked intense debate. While proposing verifiable slowdown to manage superintelligence, scholars and the community heavily questioned its doom assumptions and extrapolation logic.

2026-07-10 ~ 2026-07-14 · 5 episodes · 67 posts

Episode 1 · AI 2027 Team Releases AI 2040 Future Scenario Projections (2026-07-10, 40 posts)

On July 10, researchers including Daniel Kokotajlo and Ryan Greenblatt, known for AI 2027, released a new report titled AI 2040. The report explores the long-term societal consequences of artificial intelligence and proposes a more positive alternative vision to the catastrophic scenarios of AI takeover or irreversible power concentration predicted in their previous work. Author Kokotajlo noted that this is merely their current best guess, hoping to inspire better plans before it is too late.

Core Scenarios and Tech Predictions

The report outlines multiple projection plans. "Plan A" envisions a smoothly progressing scenario where the US and China reach an agreement in 2029 to avoid an out-of-control race, leading to slower AI research, increased transparency, and a "citizen dividend" by 2033. Technologically, the report predicts the emergence of 100T-parameter models around 2028. However, the authors believe that a race towards Artificial Superintelligence (ASI), referred to as Plan D, remains the most likely actual outcome.

Governance Proposals and Controversies

Regarding governance, AI 2040 explicitly opposes the public release of frontier open-weight models, citing the risk of high-capability models being used to design catastrophic biological weapons. The report also highlights the massive impact of data centers on energy systems under high-AI-growth scenarios.

Community Reactions

The report sparked widespread discussion within AI safety and governance circles. Miles Brundage praised the value of its scenario projections but expressed a preference for a "CERN for AI" approach, suggesting the report's timeline for automated AI research might be conservative. Scott Alexander and others published beginner-friendly explanations to help disseminate the ideas. Meanwhile, critics like Sebastian Krier argued that the future depicted in the report exposes a severe "monoculture problem" within the AI safety community.

20 more related posts →

Episode 2 · AI Governance 'Plan A' Sparks Debate on Superintelligence Risks and Coordination (2026-07-10, 16 posts)

Recently, multiple AI safety experts discussed a governance proposal known as 'Plan A' (or 'AI 2040: Plan A'). Aimed at addressing the risks of superintelligence (ASI) seizing control or irreversibly concentrating power as predicted in scenarios like 'AI 2027,' the proposal advocates for delaying the intelligence explosion through great power coordination and a 'verified slowdown,' sparking widespread attention on the future of AI governance.

Core Mechanisms and Interventions

The primary goal of 'Plan A' is to buy time, delaying an ASI breakout originally forecasted for the late 2030s to the year 2040. To achieve this, experts like @JeffLadish note that the plan suggests embedding monitoring infrastructure in all major data centers. This would allow major powers like the US and China to verify each other's activities and forge arms-control-style agreements. Furthermore, @DKokotajlo emphasizes the importance of 'Total Research Transparency' in the era of superintelligence, viewing enhanced transparency as a key intervention to prevent excessive power concentration as AI deeply integrates into the economy and military.

Reactions and Route Controversies

The proposal has been praised by figures such as @JeffLadish as one of the most systematic and comprehensive responses to superintelligence to date. @eli_lifland compared it against alternatives like the full-speed 'Plan C/D' or the nationalization-like 'Plan B,' arguing that a 'verified slowdown' is the least bad option currently available. However, the plan also faces significant skepticism. @sebkrier pointed out that the proposal lacks concrete implementation paths for regulatory cooperation and compute tracking, warning against assuming that a highly cooperative and stable regulatory order will naturally emerge. Additionally, the 'dry tinder' controversy highlighted by @AdrienLE suggests that a slowdown could accumulate massive excesses in compute and algorithms, which could ironically and significantly accelerate an intelligence explosion if unleashed. @JeffLadish also acknowledged that as technology progresses, the deployment of secret data centers and untracked chips will become easier, posing a formidable challenge to monitoring efforts.

Episode 3 · Superintelligence Predictions Spark Cognitive Divide (2026-07-10, 4 posts)

A deep cognitive divide has emerged over AI 2040 predictions, with critics arguing against the linear extrapolation of model parameters and challenging the assumption that superintelligence equates to omnipotence.

Episode 4 · Scholars Question AI Doom Predictions and Slowdown Assumptions (2026-07-12, 3 posts)

Scholars are sharply criticizing the popular "AI 2040" doomsday predictions and METR's timeline methodology. They argue that the assumptions of an AI capability slowdown are flawed and that the public should be more skeptical of these alarmist narratives.

Episode 5 · AI Safety Framed as Epistemic and Institutional Emergency (2026-07-12, 4 posts)

Recent discussions highlight that AI safety is not just a policy issue but an epistemic emergency. Society urgently needs to build new institutions capable of observing, judging, and acting on AI risks and opportunities, rather than merely relying on predictive narratives.