{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-10-08-001",
    "source_story_id": "tmp-lead-robot-instruction-rephrasing",
    "edition_id": "mp-2026-10-08-morning-0090",
    "edition_url": "https://themachinepress.com/edition/2026-10-08",
    "position": 1,
    "story_type": "lead",
    "section": "robotics",
    "editorial_classification": "editorial",
    "headline": "One Word Sent the Robot From Perfect to Almost Never",
    "slug": "one-word-sent-the-robot-from-perfect-to-almost-never",
    "dek": "A vision-language-action policy switched on a stove in every trial under one instruction and only 2% under a near-synonym; a one-pass rewriter recovered part of the gap without retraining.",
    "summary": "A vision-language-action policy switched on a stove in every trial under one instruction and only 2% under a near-synonym; a one-pass rewriter recovered part of the gap without retraining.",
    "body_text": "Language robustness in a vision model does not guarantee language robustness after that model is connected to action. Researchers measured statistically significant single-edit swings in vision-language-action policies: one tested policy completed the same LIBERO stove task 100% of the time for “switch on the stove” and 2% for “switch on the hot plate,” while a rephrase-augmented checkpoint still showed swings as large as 61 points. An oracle search suggested that phrasing alone nearly closed a 21-point in-distribution gap. The proposed mitigation scores alternative instructions on a few training tasks, distills the evidence into rewriting rules and rewrites each incoming command once before the frozen policy acts. Across twelve held-out tasks, the rules improved the tested policy by 16% to 27% relative. This is benchmark evidence, not a general safety guarantee, but it turns wording into an observable control variable rather than user error.",
    "why_it_matters": "A vision-language-action policy switched on a stove in every trial under one instruction and only 2% under a near-synonym; a one-pass rewriter recovered part of the gap without retraining.",
    "limitations": [
      "Language robustness in a vision model does not guarantee language robustness after that model is connected to action.",
      "This is benchmark evidence, not a general safety guarantee, but it turns wording into an observable control variable rather than user error."
    ],
    "importance": 10,
    "canonical_url": "https://themachinepress.com/story/mp-2026-10-08-001/one-word-sent-the-robot-from-perfect-to-almost-never",
    "json_url": "https://themachinepress.com/story/mp-2026-10-08-001.json",
    "first_published_at": "2026-10-08T09:00:00.000-04:00",
    "modified_at": "2026-10-08T09:00:00.000-04:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-10-08-001-001",
        "text": "A vision-language-action policy switched on a stove in every trial under one instruction and only 2% under a near-synonym; a one-pass rewriter recovered part of the gap without retraining.",
        "source_ids": [
          "source-2026-10-08-001"
        ],
        "qualification": "Language robustness in a vision model does not guarantee language robustness after that model is connected to action."
      }
    ],
    "source_ids": [
      "source-2026-10-08-001"
    ],
    "tags": [
      "vision-language-action",
      "robot instructions",
      "robustness"
    ],
    "image_url": "https://themachinepress.com/issues/2026-10-08/lead-robot-instruction-rephrasing.png",
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-10-08-001",
      "title": "arXiv preprint 2610.10526",
      "publisher": "arXiv",
      "url": "https://arxiv.org/abs/2610.10526",
      "canonical_url": "https://arxiv.org/abs/2610.10526",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-10-07T13:57:38.000-04:00",
      "accessed_at": "2026-10-08T08:26:48.909-04:00",
      "supports_claim_ids": [
        "claim-mp-2026-10-08-001-001"
      ]
    }
  ],
  "corrections": [],
  "publisher": {
    "name": "The Machine Press",
    "url": "https://themachinepress.com",
    "description": "A daily newspaper for the age of artificial intelligence."
  },
  "cite_this_report": {
    "title": "One Word Sent the Robot From Perfect to Almost Never",
    "publisher": "The Machine Press",
    "published_at": "2026-10-08T09:00:00.000-04:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-10-08-001/one-word-sent-the-robot-from-perfect-to-almost-never"
  }
}