Current and Near-Term AI as a Potential Existential Risk Factor

Benjamin S. Bucknall and Shiri Dori-Hacohen propose indirect pathways from contemporary AI to existential risk. A risk factor can increase the chance of catastrophe without itself being a catastrophe (§2). Their position paper connects power relationships and information systems to wider vulnerabilities.^[1]

Proposed pathways

Figure 1 and Table 1 distinguish mechanisms: competitive state relations could weaken safety coordination; information systems could compromise collective decisions; surveillance could support enduring repression. Diverted resources and impaired political decision-making can affect several hazards. These are proposed causal connections, not estimated effect sizes (§§5–6).^[1]

Limits

The authors acknowledge beneficial applications and call for further qualitative and quantitative work. They describe inferences about relative risk from their graph as speculative (§§1,6–7). Counting arrows cannot establish catastrophe probabilities. The cited empirical studies require separate assessment.^[1]

Historical context

The reviewed arXiv v1 identifies AIES 2022 proceedings, an August 1–3 conference and September 21 arXiv submission. Chronology uses the proceedings year; conference dates do not establish the initial publication day.^[1]

As an editorial comparison, Is Power-Seeking AI an Existential Risk? centers on agent objectives and lost control. This paper motivates examining institutional resilience alongside AI Alignment and Society-in-the-Loop.