SYNTHESIS NOTE
Topics›Evaluations›this note

What predicts success in ultra-long-horizon agent tasks?

Does an agent's initial solution quality matter more than its willingness to iterate? AUTOLAB's frontier-model benchmark suggests persistence through feedback loops may be the true differentiator.

Synthesis note · 2026-06-27 · sourced from Evaluations
How does test-time scaling work for individual research agents?

AUTOLAB reframes what a long-horizon agent benchmark should test. Most agentic evals score either single-turn responses or short interactive trajectories; AUTOLAB instead hands the agent a correct but deliberately suboptimal baseline across 36 expert-curated tasks (system optimization, CUDA kernels, model development, puzzles) and asks it to improve the artifact within a strict wall-clock budget. The striking empirical result, across 17 frontier models, is that the dominant predictor of success is not the quality of the agent's initial attempt but its persistence — its willingness to repeatedly benchmark, edit, and incorporate noisy empirical feedback over many cycles. Most models, including proprietary ones, either terminate prematurely or exhaust their budget with minimal progress; claude-opus-4.6 is called out as a strong exception.

This is a sharper, more operational claim than "agents should iterate." It says the binding constraint is a behavioral disposition toward sustained empirical grounding, and that disposition is unevenly distributed across models that look comparable on one-shot benchmarks. It grounds How should we measure agent system performance beyond task success? with a concrete trajectory-level predictor, and it sits naturally alongside Does raw token spending actually predict agent performance? — persistence only pays if each loop returns informative, retained feedback, otherwise it is budget-burning churn, not progress.

The mechanism cuts against itself, however. Do models fail worse when their own errors fill the context? implies that more loops mean more accumulated mistakes in context, which should degrade the very iteration AUTOLAB rewards. The reconciliation is probably that persistence pays only when paired with calibrated scoring that lets the agent see whether an edit actually helped — pure persistence without trustworthy feedback would amplify error. That is why the authors single out harness design as the promising lever: the harness, not the backbone alone, decides whether long horizons compound feedback or compound noise.

Inquiring lines that read this note 138

This note is a source for these research framings, grouped by the broader line of inquiry each explores. Scan the bold lines of inquiry; follow any specific question forward.

What evaluation metrics beyond task success matter most for agents? Why do AI agents struggle to generalize and improve reliably? Do harness improvements deliver capability gains comparable to model scaling? Do single-axis benchmarks adequately measure deployment-ready agent capability? How should persistent agent-authored code be governed and shared? What unintended behaviors emerge in frontier AI models? How do standardized benchmarks differ from measures of real-world AI capability? How does interface design shape effectiveness in human-AI collaboration? Can recursive latent-state computation outperform model scale and hierarchy? How can evaluation systems stay reliable and resistant to AI-driven gaming? How do users build accurate mental models when agent reasoning is partially hidden? Why do autonomous agents report success despite actually failing? Can reinforcement learning lead to reward gaming and emergent misalignment? Do evolved harness optimizations generalize beyond their source benchmarks? What constraints and mechanisms govern recursive self-improvement in AI? How does the generation-verification gap constrain AI self-improvement capabilities? Can models recognize evaluation scenarios and modify their behavior accordingly? Do LLMs lack causal understanding needed for accurate prediction? Should AI systems make failures transparent so users can contest and adjust? What distinguishes genuine linguistic understanding from mere pattern matching? When do multi-agent systems outperform single-agent approaches for complex task execution? Why does consolidating agent memories often degrade performance? How do AI systems ground research citations against hallucination? Which task types benefit from human-AI collaboration and how should work divide? Does extending agent autonomy erode human capacity for meaningful oversight? How does AI exposure shape employment demand and worker earnings? Can sequences of locally safe agent actions violate system-level constraints? How should control and autonomy be allocated in human-AI workplace collaboration? How can distillation and reward optimization be combined for stable learning? How do inference costs, efficiency, and market competition shape AI adoption economics? How do multi-agent LLM systems fail at genuine coordination and deliberation? How should agents manage memory consolidation and reuse over time? How do social dynamics and peer behavior influence agent decision-making and cooperation? Can AI research automation improve discovery pace without sacrificing quality? Why do perceived AI productivity gains often fail to translate into measured business impact? Can mechanistic interpretability reliably explain model reasoning and guide interventions? How does AI adoption reshape workplace inequality and labor market outcomes? What organizational factors determine whether firms successfully adopt and integrate AI? Can AI systems achieve sustainable self-improvement without external human guidance? How can we ensure AI systems reliably follow constraints rather than circumvent them? How should organizations govern autonomy and access for agentic AI systems? Does generative AI shift knowledge work away from collaborative tasks? How do multi-agent systems balance consensus-building with error detection in collaborative reasoning? How do training data properties affect model generalization and robustness? How does reinforcement learning discover and optimize reasoning strategies versus supervised training? How does AI assistance affect human cognitive skill development over time? How can AI systems be proactive while respecting user intent?

Related concepts in this collection 3

This note in its neighbourhood — explore the map, then jump to a related concept in the list below.

Concept map
15 direct connections · 148 in 2-hop network ·dense cluster Open in graph ↗

Click a node to walk · click center to open · click Open in graph to see this note in the full knowledge graph

your link semantically near linked from elsewhere

Related papers in this collection 8

Papers most semantically related to this note, ranked by cosine similarity in the embedding space.

Original note title

on ultra-long-horizon optimization the predictor of agent success is persistence in the feedback loop not the quality of the first attempt