AI is a dangerous risk we should take

The author argues that rapid AI development, despite uncertain catastrophic risks—including one expert’s estimate of a 10% chance AI kills all humans within the next decade—could reduce humanity’s broader systemic risks. Proposed benefits include making bank resolution more feasible, improving early pathogen detection and countermeasures, and supporting a self-sustaining civilization beyond Earth, a project the author says would take decades. This is an opinion article advocating against a global AI pause, not a report of a new policy or market event.

Analysis

The investment gap is between AI’s potential social value and its near-term ability to generate attributable cash flows. Better tools for bank resolution or pathogen surveillance could lower tail risk, but neither benefit is presently a reliable revenue stream for META or a measurable reduction in UBS’s resolution burden. Treat the essay as a policy argument, not an earnings catalyst.

For UBS, the second-order risk runs both ways: more capable supervisory and resolution tools could make authorities more willing to impose losses on creditors and shareholders, weakening the implicit support value attached to systemically important banks. Conversely, more effective crisis management could reduce disorderly contagion and the likelihood of a forced rescue. The net effect on UBS funding costs and equity value is unproven; resolution rules, not model capability alone, determine outcomes.

META’s exposure is indirect. A broad AI pause or tighter model-development rules could constrain its product roadmap, while permissive policy supports investment and competition. The article offers no company-specific evidence to revise estimates. The contrarian point is that “AI as a hedge” is not necessarily an investable thesis: benefits are diffuse, long-dated, and may be captured by governments or users, while development costs, competitive intensity, and misuse risks accrue sooner. Over 1–3 months, watch legislative and regulatory action; over 6–18 months, watch whether AI deployment produces auditable productivity gains or changes bank resolution requirements. The article alone does not justify a directional position.

Key Decisions for Investors

• No trade from this article alone. Do not translate the societal case for faster AI progress into a near-term META earnings upgrade without evidence of monetization, adoption, or guidance changes.
• For UBS, monitor resolution-plan reviews, regulatory capital/TLAC requirements, and wholesale funding spreads. Reassess the risk if authorities explicitly cite AI-enabled resolvability when changing resolution or support policy.
• Use AI-pause legislation and major model-development restrictions as a 1–3 month catalyst watch for META’s regulatory-risk assessment; distinguish proposed bills from enacted rules and implementation timelines.
• For a 6–18 month thesis check, seek independently verifiable evidence that AI reduces resolution time/cost or improves supervisory outcomes. If such evidence does not emerge, treat claimed systemic-risk reduction as non-monetizable rather than a valuation input.