{"id":"4459b66a-3919-4216-a0f1-9b82cb30c047","arxiv_id":"2606.12428","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":5.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":0,"one_line_summary":"A dynamic scraping tool and analysis of 66 AI majors and 87 AI minors across US universities reveals high variability in program size and requirements, including that all majors without a general AI course require machine learning and that ethics courses appear in over a third of majors but under a ","lead":"This paper reports on the state of undergraduate AI programs in the US as of Spring 2026 using a scraping tool that maps over 350 programs at 560 institutions representing 86% of CS graduates. A smart generalist might read it to understand current variability in AI major and minor requirements and access a public tool for exploring them.","discovery_kind":"new_application","skeptic_critique":{"model":"grok-4.3","headline":"Unvalidated scraping tool accuracy is the load-bearing risk for the reported course requirement percentages","rationale":"The reader’s weakest_assumption is precisely the single point on which the empirical takeaways depend. No other internal inconsistency or derivation error is visible in the described analysis; the concern is purely about data fidelity, which the proposed audit would resolve.","tokens_in":1878,"tokens_out":341,"duration_ms":11834,"concrete_test":"Draw a random sample of 20 majors and 20 minors from the reported set; for each program, have two independent human coders extract the exact required-course list from the current university catalog and code presence of general AI, ML, and Ethics courses. Compare to the tool output; if disagreement rate on any of the three categories exceeds 15% or if any program is missing from the tool, recompute the two takeaways on the audited subset.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The two takeaways are quantitative claims: (1) every major without a general AI course requires ML instead, and (2) >1/3 of majors but <1/4 of minors require an Ethics in AI course. These rest on the tool having correctly classified course presence/absence across the 66 majors and 87 minors extracted from 560 institutions. The manuscript describes the scraping process and the cicmap.ai interface but reports no manual validation, precision/recall figures, or audit of a sample against source catalogs. Catalog pages vary widely in structure, terminology, and update frequency; any systematic mis-mapping of “AI”, “ML”, or “Ethics” labels would directly alter the cited fractions.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.3","summary":"The paper describes a web-scraping tool and public interface (cicmap.ai) that searched 560 U.S. four-year institutions (covering 86% of CS graduates) to identify more than 350 undergraduate AI programs (majors, minors, concentrations, certificates) as of early 2026. It supplies a snapshot of these programs and analyzes course requirements across 66 AI majors and 87 AI minors, reporting high variability in program size and two quantitative findings: every major without a general AI course requires an ML course instead, and more than one-third of majors but fewer than one-quarter of minors require an Ethics in AI course.","tokens_in":2028,"tokens_out":410,"duration_ms":16738,"significance":"If the data extraction proves reliable, the work supplies a timely public resource and historic record of AI education during rapid growth, with the continually updating cicmap.ai tool as a concrete contribution that could aid students, advisors, and administrators. The scale of the institutional sample is a strength.","major_comments":[{"comment":"The description of the scraping process and cicmap.ai interface (abstract and methods) reports no validation of detection, classification, or extraction accuracy—no precision/recall figures, no inter-rater reliability, no manual audit of a sample of catalogs against source pages, and no discussion of deduplication or filtering rules. This directly undermines the two central quantitative claims in the analysis of majors and minors, which rest on correct identification of AI, ML, and Ethics course presence/absence across all 66 majors and 87 minors.","section":"Methods / scraping tool description and analysis of majors/minors"}],"minor_comments":[{"comment":"The abstract states the tool 'dynamically update[s]' but provides no details on update frequency, handling of catalog changes, or how users can verify current data.","section":"Abstract"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":{"model":"grok-4.3","summary":"We thank the referee for their constructive feedback and positive assessment of the work's timeliness, scale, and potential utility as a public resource. The major comment identifies a clear gap in the methods description regarding validation and procedural details, which we will address in revision. Below we respond point by point.","responses":[{"response":"We agree this is a substantive omission. The manuscript does not report validation metrics or detail the deduplication and filtering rules, which weakens confidence in the reported counts and the two quantitative takeaways. In the revised manuscript we will add a new Methods subsection that (a) explicitly describes the deduplication logic (unique keys combining institution, program title, and catalog URL) and filtering criteria (undergraduate-only, AI-focused programs at four-year institutions), and (b) presents results from a manual audit of a random sample of 50 institutions. The audit will compare tool outputs against direct catalog inspection and report precision, recall, and F1 scores for both program detection and course-type classification. If inter-rater checks were performed during tool development they will be reported; otherwise we will note that limitation and describe the single-reviewer protocol used. These additions will directly support the reliability of the 66-major / 87-minor analysis.","revision_made":"yes","referee_comment":"[Methods / scraping tool description and analysis of majors/minors] The description of the scraping process and cicmap.ai interface (abstract and methods) reports no validation of detection, classification, or extraction accuracy—no precision/recall figures, no inter-rater reliability, no manual audit of a sample of catalogs against source pages, and no discussion of deduplication or filtering rules. This directly undermines the two central quantitative claims in the analysis of majors and minors, which rest on correct identification of AI, ML, and Ethics course presence/absence across all 66 majors and 87 minors."}],"tokens_in":1499,"tokens_out":406,"duration_ms":34001,"standing_objections":[]},"desk_editor":{"model":"grok-4.3","letter":"The paper delivers new counts and breakdowns for 66 AI majors and 87 minors plus a live explorer at cicmap.ai that covers 86% of US CS graduates. That is the core value: a current record and searchable resource during fast program growth.\n\nIt does the collection work at scale by searching 560 institutions and extracting requirement details that prior surveys did not report. The two takeaways—no general AI course without an ML substitute, and ethics requirements in over a third of majors but under a quarter of minors—are the kind of concrete patterns that education stakeholders can use right away.\n\nThe soft spot is exactly where the stress test flagged: no validation of the scraper. The manuscript describes the process but gives no precision or recall numbers, no inter-rater checks, and no sample audit against official catalogs. University pages vary in structure and terminology, so systematic misreads on “AI,” “ML,” or “Ethics” labels would shift the reported fractions. That is a load-bearing gap for the quantitative claims.\n\nThis is for department chairs, advisors, and students who need to see how AI degrees are actually structured today. A reader who wants current enrollment or curriculum data will get direct use from the tool and tables. It deserves peer review because the scale of the collection is substantial and the public interface is a real output, even if the methods section needs strengthening on accuracy before the numbers can be treated as reliable.","headline":"A practical snapshot and public tool for tracking US AI undergrad programs, but the course requirement takeaways rest on unvalidated scraping with no accuracy checks.","tokens_in":2530,"tokens_out":360,"would_cite":false,"duration_ms":15824,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.3","headline":"A 2026 mapping of US undergraduate AI programs finds high variability in requirements, with majors needing either a general AI or ML course and ethics included in more than a third of majors versus under a quarter of minors.","keywords":["AI education","undergraduate programs","AI majors","AI minors","machine learning requirements","ethics in AI","US universities","program mapping"],"falsifier":"A manual audit of university websites that finds multiple AI programs or specific course requirements the tool missed or misclassified.","tokens_in":2790,"feed_emoji":"🗺️","tokens_out":699,"duration_ms":27581,"temperature":0.7,"pith_summary":"The paper creates a snapshot and dynamic tool to track undergraduate AI programs across more than 560 US institutions that account for 86 percent of computer science graduates. It identifies over 350 programs including majors, minors, concentrations, and certificates. Analysis of 66 majors and 87 minors documents wide differences in program size and required courses. Two consistent patterns emerge: programs without a general AI course instead require machine learning, and ethics courses appear in over a third of majors but fewer than a quarter of minors. This record matters for students, advisors, and faculty navigating the expansion of AI degrees.","feed_headline":"Most US AI majors require either AI or ML courses","feed_subtitle":"2026 analysis of 66 majors and 87 minors finds ethics required in over a third of majors but under a quarter of minors, with wide variabilit","key_machinery":"The scraping and mapping tool that detects, extracts, and displays course requirements from university websites to produce an updatable inventory of AI programs.","core_discovery":"The paper establishes a historic record of AI programs in the United States in early 2026 through systematic scraping of university websites, locating more than 350 undergraduate programs at 560 institutions. Analysis of the 66 majors and 87 minors shows substantial variation in the number and type of required courses, but reveals that every major requires either a general AI course or a machine learning course and that ethics in AI is mandated in more than a third of majors yet in just under a quarter of minors.","pith_inferences":["The observed differences suggest that students completing only a minor may receive less exposure to AI ethics than those completing a major.","Continued growth in new programs will require ongoing updates to keep the inventory current.","The high coverage of institutions producing most CS graduates makes the patterns representative for the majority of computing students.","Variability in requirements could prompt discussion of whether some standardization of core AI and ethics content would benefit students transferring between institutions."],"forward_implications":["AI majors and minors differ substantially in total credit hours and specific course mandates across institutions.","Machine learning functions as the required substitute whenever a general AI course is absent from a major.","Ethics in AI appears as a required course more often in majors than in minors.","The inventory supplies a continually updated resource for comparing program requirements."],"fun_headline_variants":["All US AI Majors Require AI or ML Course","2026 Survey Maps 350 AI Programs at 560 Schools","Ethics Mandated in 1/3 of Majors but 1/4 of Minors","AI Degrees Show High Variability in Required Courses","Every Major Needs AI or ML; Ethics Rules Differ"],"cache_read_input_tokens":2112,"weakest_assumption_plain":"The scraping tool correctly detects, classifies, and extracts accurate course requirements from university websites for all relevant programs without significant omissions, misclassifications, or outdated information across the 560 institutions sampled.","fun_headline_variants_meta":{"raw":{"variants":["All US AI Majors Require AI or ML Course","2026 Survey Maps 350 AI Programs at 560 Schools","Ethics Mandated in 1/3 of Majors but 1/4 of Minors","AI Degrees Show High Variability in Required Courses","Every Major Needs AI or ML; Ethics Rules Differ"]},"model":"grok-4.3","cost_usd":0.004311,"raw_usage":{"total_tokens":2239,"prompt_tokens":814,"num_sources_used":0,"completion_tokens":81,"cost_in_usd_ticks":43112000,"prompt_tokens_details":{"text_tokens":814,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":1344,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":814,"tokens_out":81,"duration_ms":15222,"temperature":1.0,"reasoning_tokens":1344,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-06-30T19:49:56.068827+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"A manual audit of university websites that finds multiple AI programs or specific course requirements the tool missed or misclassified.","supporting_citations":[],"review_version":1}