New nonprofit STABL launches to study societal bottlenecks of advanced AI
STABL will focus on empirical research about AI progress, recursive self-improvement, and institutional resilience amid rapid AI diffusion.
1 source · cross-referenced
- A new nonprofit research lab, STABL, was announced to study the societal impacts and bottlenecks of advanced AI.
- The lab argues that institutional and organizational change—not just AI capabilities—will determine the pace of societal transformation.
- Research priorities include measuring recursive self-improvement (RSI), evaluating open-ended AI capabilities, and assessing downstream societal outcomes of AI adoption.
- STABL positions itself in the "top-right quadrant" of a 2x2 framework balancing AI’s transformative potential with attention to bottlenecks.
A new nonprofit research organization, STABL, was launched to study the societal transformation and bottlenecks associated with advanced AI. The lab’s founders argue that the pace of societal change will be constrained more by institutional and organizational adaptation than by raw AI capabilities. Their work emphasizes the need for evidence-based research on AI progress, including the prospects of recursive self-improvement (RSI) and the bottlenecks that limit how AI capabilities translate into societal outcomes.
STABL frames its worldview on a 2x2 chart, placing itself in the top-right quadrant. This quadrant balances two perspectives: taking AI’s transformative potential seriously while also acknowledging the importance of downstream bottlenecks—such as regulatory, organizational, and governance constraints—that shape AI’s real-world impact. The lab contrasts this with other quadrants: the top-left, which reflects technological determinism (e.g., claims that superintelligence will unilaterally transform society), and the bottom-right, which prioritizes preserving existing institutions at the expense of technological progress.
The lab’s research agenda includes empirical studies on the pace of AI progress and RSI, as well as analyses of risk mitigation strategies beyond model alignment and non-proliferation. STABL also plans to measure the specific benefits and harms of AI adoption, study labor market disruptions, and propose ways to improve societal resilience and revitalize institutions. These priorities reflect a focus on bridging gaps between AI’s potential and the practical constraints of its diffusion.
One of STABL’s key research directions involves measuring progress toward RSI, a concept central to many forecasts of explosive AI advancement. The lab plans to use open-ended evaluations, such as the CRUX framework, to probe the upper bounds of AI capabilities on long, complex tasks—including whether AI agents can conduct open-ended AI research. Initial results from CRUX suggest that current frontier agents cannot yet perform such tasks, as expert reviewers rejected papers produced by these systems. STABL intends to repeat these evaluations regularly as models and scaffolds improve.
STABL’s founders emphasize the need for intellectual diversity in AI risk research, noting that much of the current literature on RSI is dominated by assumptions aligned with technological determinism. They argue that a broader ecosystem of researchers and organizations—with varied priors about AI progress—could test these assumptions more rigorously, for example by comparing causal models of AI progress rather than relying on a single contested model.
- Nov 11, 2026 · Wired
EU proposes Digital Fairness Act to ban dark patterns and restrict addictive design features
Trust75 - Nov 10, 2026 · TechCrunch — AI
New York enforces AI safety law requiring registration and incident reporting for frontier labs
Trust75 - Nov 10, 2026 · Import AI — Jack Clark
New Realtime-Bench shows frontier AI models struggle with low-latency tasks
Trust75