AI Futures Project releases normative 2040 scenario blueprint

The AI Futures Project has published AI 2040: Plan A as a normative scenario for how society should approach a future with many powerful AI systems. The document represents a deliberate shift from the group's earlier AI 2027 work, which was framed as a forecast — an attempt to describe what the authors believed such a future would actually look like. AI 2040, by contrast, is explicitly normative: it describes what ought to be done, assuming that a future with multiple powerful AI systems is likely.

The plan includes what the authors call an "incremental AI policy wishlist" — a set of policy recommendations intended to be executed in the near term. One notable element focuses on limiting the intelligence gap between internal and external model deployments. In other words, the plan recommends efforts to narrow the difference in capability between AI systems that organizations use internally and those they deploy externally to the public. This recommendation reflects an underlying concern about asymmetries in AI access and control.

Early reviewers have identified substantive objections to the proposal on multiple grounds. Some objections are "within-frame" — meaning they accept the plan's stated goals but question whether the proposed steps would actually achieve them. Others reflect external value disagreements — concerns that pursuing AI 2040 would sacrifice or harm things the reviewers consider important, independent of the plan's internal coherence. The nature and weight of these objections remain under discussion within the AI policy and research community.

The AI Futures Project's release of AI 2040: Plan A marks a shift from forecasting to prescription. Where AI 2027 aimed to describe a plausible future with advanced AI systems, AI 2040 moves into the normative domain — arguing for specific actions policymakers and organizations should take now to navigate that future responsibly. The plan reflects growing concern in AI governance circles about the pace and distribution of AI capabilities. One of its core recommendations — managing the intelligence gap between internal and external deployments — reflects a concern that organizations may retain more powerful AI systems for internal use while deploying less capable versions externally, potentially widening asymmetries in capability and control.

The plan has already drawn critical attention. Reviewers have raised both internal-consistency objections (questioning whether the proposed steps would actually achieve the authors' stated goals) and substantive disagreements about whether pursuing the plan's recommendations would preserve other values observers care about. This suggests the proposal is being engaged seriously by the AI policy community, even as its merits remain contested.

No comments yet. Be the first to share your thoughts!