{ "schema": "cortist.milestone-candidates.v1", "published": "2026-10-27", "candidateCount": 5, "probabilitiesIssued": 0, "lockedForecasts": 0, "scoredForecasts": 0, "status": "Research watchlist only; no numerical forecast published", "scoring": "Only a separately locked, timestamped and independently resolvable binary forecast may be scored via (p-outcome)^2. Never retrospectively score this candidate register.", "criteria": [ "Fixed question and time window", "Official or independently auditable evidence", "Baseline excludes already-achieved events", "Published and frozen event definition before numerical estimate", "Public immutable version with proof of timestamp", "Prospective independent resolution and proper scoring" ], "items": [ { "id": "CM-001", "slug": "private-coding-agent-90-percent-2027", "topic": "AI Agents", "topicSlug": "ai-agents", "question": "Will an AI coding agent resolve 90% of SWE-bench Pro V2 private tasks by October 2027?", "metric": "SWE-bench Pro V2 private benchmark: at least 90% audited pass@1", "deadline": "2027-10-27", "baseline": "Scale Labs reported 222/272 private tasks resolved for a leading agent in September 2026; near-perfect public scores are not comparable.", "yesRule": "The benchmark owner publishes a qualifying audited result on the frozen private V2 task set no later than the deadline.", "noRule": "No qualifying result after an official dated archive review.", "exclusions": "Public scores, self-reported results, changed task mixes and different test versions do not count.", "prelockBlocker": "Freeze a permanent private-split identifier, task denominator and official archive before probability issuance.", "sources": [ { "title": "Primary source / methodology", "url": "https://labs.scale.com/blog/swe-bench-pro-v2" }, { "title": "Official reference / evidence", "url": "https://labs.scale.com/leaderboard/swe_bench_pro_public_v2" } ], "rationale": "Strongest because withheld tasks reduce contamination risk; not a test of general intelligence.", "state": "research-draft", "probability": null, "scorable": false, "publishedCandidate": "2026-10-27" }, { "id": "CM-002", "slug": "ladder-five-object-robot-2027", "topic": "AI & Robotics", "topicSlug": "ai-and-robotics", "question": "Will a robot complete at least 50% of five-object LADDER tasks by October 2027?", "metric": "Real-robot LADDER five-object sequences: at least 50% complete", "deadline": "2027-10-27", "baseline": "INSAIT introduced LADDER on 26 October 2026 and reported persistent difficulty with long sequences.", "yesRule": "A benchmark-owner or independent evaluator publishes 50% or higher end-to-end real-robot success on a pre-registered identical five-object task subset.", "noRule": "A complete dated archive of the agreed task split shows less than 50% at deadline.", "exclusions": "Simulation-only scores, teleoperated assistance, single-object tasks and selectively edited demos do not count.", "prelockBlocker": "Freeze exact five-object subset, trial counts, evaluation protocol and result archival before locking.", "sources": [ { "title": "Primary source / methodology", "url": "https://insait.ai/ladder-a-new-benchmark-for-evaluating-ai-on-complex-robot-tasks/" }, { "title": "Official reference / evidence", "url": "https://ladderrobotics.ai/" } ], "rationale": "Direct physical reliability test; definition not yet ready to score.", "state": "research-draft", "probability": null, "scorable": false, "publishedCandidate": "2026-10-27" }, { "id": "CM-003", "slug": "first-eu-gpai-article-101-fine-2027", "topic": "AI & Governance", "topicSlug": "ai-and-governance", "question": "Will the European Commission impose an Article 101 GPAI fine by October 2027?", "metric": "At least one formally adopted fine decision under Article 101", "deadline": "2027-10-27", "baseline": "Article 101 allows sanctions against general-purpose AI model providers; an independently frozen first-fine baseline is still required.", "yesRule": "The Commission formally adopts an Article 101 fine between 27 October 2026 and 27 October 2027 inclusive, evidenced by an official decision or register.", "noRule": "No qualifying decision after review of the official legal decisions record and publication delays.", "exclusions": "National fines, warnings, letters, requests for information or draft sanctions do not qualify.", "prelockBlocker": "Independently check baseline for earlier decisions and decide how to handle adoption versus publication timing.", "sources": [ { "title": "Primary source / methodology", "url": "https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-101" }, { "title": "Official reference / evidence", "url": "https://digital-strategy.ec.europa.eu/en/policies/enforcement-ai-act" } ], "rationale": "Enforcement is more verifiable than policy promises.", "state": "research-draft", "probability": null, "scorable": false, "publishedCandidate": "2026-10-27" }, { "id": "CM-004", "slug": "waymo-munich-paid-driverless-rides-2027", "topic": "AI & Robotics", "topicSlug": "ai-and-robotics", "question": "Will paid, driverless Waymo rides launch in Munich by October 2027?", "metric": "First publicly available paid driverless ride within Munich", "deadline": "2027-10-27", "baseline": "Waymo said in August 2026 it was mapping Munich with safety drivers and targeted public service toward the end of 2027.", "yesRule": "Waymo or the responsible regulator confirms that members of the public can book and pay for a Munich passenger ride with no in-vehicle safety driver by deadline.", "noRule": "No such service in Munich before deadline after operator and regulator records are checked.", "exclusions": "Mapping drives, safety-driver tests, only free invitation pilots or trips outside Munich do not count.", "prelockBlocker": "Freeze definition of Munich service area and evidence standard for paid publicly available service.", "sources": [ { "title": "Primary source / methodology", "url": "https://waymo-prod.appspot.com/blog/2026/08/waymo-in-munich/" }, { "title": "Official reference / evidence", "url": "https://waymo.com/blog/2025/12/2025-year-in-review/" } ], "rationale": "Commercial deployment, not a prototype announcement.", "state": "research-draft", "probability": null, "scorable": false, "publishedCandidate": "2026-10-27" }, { "id": "CM-005", "slug": "mlperf-inference-efficiency-double-2027", "topic": "AI Chips & Compute", "topicSlug": "ai-chips-and-compute", "question": "Will verified AI inference performance per watt double by October 2027?", "metric": "MLPerf Endpoints: at least 2x GPT-OSS-120B standardized throughput per watt", "deadline": "2027-10-27", "baseline": "MLCommons launched continuously reviewed Endpoints results in 2026; cross-vintage power normalization still requires a stable baseline.", "yesRule": "MLCommons verifies an Available, Standardized GPT-OSS-120B system with at least twice the pre-registered 2026 reference output tokens per watt at matching latency and quality.", "noRule": "Official comparable results remain below the locked 2x threshold at cutoff.", "exclusions": "Different models, changed quality limits, vendor-only figures and Preview/RDI submissions do not count.", "prelockBlocker": "Critical: freeze 2026 baseline, comparable power normalization, model/tokenizer, quality tier, latency and reviewer-approved rules.", "sources": [ { "title": "Primary source / methodology", "url": "https://mlcommons.org/benchmarks/endpoints/" }, { "title": "Official reference / evidence", "url": "https://docs.mlcommons.org/endpoints-submission-docs/understand/what-is-mlperf-endpoints/" } ], "rationale": "Chips and inference efficiency need full-system like-for-like measurement.", "state": "research-draft", "probability": null, "scorable": false, "publishedCandidate": "2026-10-27" } ] }