Integrating AI into First-Year Research as Part of the Process, Not The Process

One of the recurring challenges in teaching legal research is helping students understand that the law is simultaneously uniform and jurisdiction-specific. Nowhere is that tension more apparent than in negligence law. Nearly every first-year law student learns the familiar elements (duty, breach, causation, and damages), but many emerge from their introductory torts course with the impression that the negligence doctrine is essentially the same everywhere.

Reflecting on the foundational research habits instilled in introductory legal writing and research courses, carrying those timeless principles forward into today’s classroom requires a modern twist: bridging traditional doctrinal precision with the realities of emerging technology. This fall, in my Lawyering Skills I (legal writing & research) course, I am testing that balance through a collaborative fifty-state negligence research project that treats Gen AI not as the sole research mechanism, but as one instrument within a comprehensive research toolkit.

What follows is a playbook and pilot proposal: designed to accomplish several pedagogical goals at once and refined in light of the tensions those goals create.

Structure & Scope

Working in groups of two, students cover five states each: five groups across ten states per team, surveying all fifty jurisdictions. To ground the project in current practice, student pairs split tools between Cocounsel and Protégé, cross-auditing how competing legal AI engines handle the exact same five-state survey. The project requires students to synthesize traditional legal research sources alongside these platforms, positioning technology as an initial stepping stone rather than the final destination.

Assignment Deliverables at a Glance (Per State)

  • Verified AI Case Leads: Students prompt an AI tool to identify leading negligence decisions across core elements (duty, breach, causation, foreseeability, damages), preserving all prompts and raw outputs. Every citation must be independently verified using traditional legal research methods. AI output is treated strictly as a hypothesis, not an answer.
  • Doctrinal Development Traced via Secondary Sources: Students consult state practice materials, legal encyclopedias, and the Restatement of Torts to evaluate how doctrine evolved and whether local courts adopted, modified, or rejected provisions from the Restatement (Second) and (Third).
  • Secondary Source Evaluation: Students locate an American Law Reports (ALR) annotation or law review article relevant to negligence in that jurisdiction, articulating why the source matters to move beyond simple citation gathering.
  • Distilled Governing Rule: Students isolate the authoritative rule statement from the state’s highest court and identify whether terminological variations (e.g., “proximate cause,” “legal cause,” “substantial factor,” “scope of liability”) represent semantic or substantive differences.
  • A Memorable Anchor Case: Students select one factually compelling decision per state that makes the underlying doctrinal principle stick through narrative.

Design Decisions & Pedagogical Tensions

Two core structural choices in this playbook warrant transparency.

1. Front-Loading AI as a “Hypothesis Engine”

Placing AI at the beginning of the research workflow rather than at the end reflects the reality that today’s students will encounter in practice. However, this creates a deliberate asymmetry: AI-generated leads carry a verification burden that traditional sources (like a Westlaw headnote) do not.

Students are expected to encounter, and document, instances in which AI platforms overstate the significance of a case, mischaracterize doctrinal developments, or confuse persuasive authority with controlling law. Working out and with these bugs is the core assignment. Framing AI outputs as unverified hypotheses ensures that technology remains part of the process: a starting hypothesis that triggers, rather than replaces, traditional analytical labor.

2. Depth vs. Breadth

Surveying five states per student trades deep jurisdictional specialization for broad comparative awareness. The payoff is immediate: students discover that while the broad negligence framework is remarkably uniform, deep splits exist in terminology, methodology, and policy. The trade-off is potential superficiality. Future iterations may calibrate this balance by having students establish deep mastery in one “anchor state” while surveying the other four more lightly.

Assessment & Resources

Assessment Realities

Because a 50-state survey generates a significant volume of student deliverables, rubrics are essential for targeted grading. Assessment evaluates three key dimensions:

  • Accuracy: Verification rigor against primary law.
  • Doctrinal Precision: Quality of rule statements synthesized against the controlling authority.
  • Comparative Depth: Ability to distinguish genuine doctrinal divergence from mere terminological variation.

Resource Prerequisites

This assignment relies on a robust hybrid research environment. UIC Law students utilize:

  • Primary Case Law Databases: Westlaw and Lexis for primary law verification.
  • Scholarly & Draft Repositories: HeinOnline (along with Westlaw and Lexis) for Restatement drafts and secondary scholarship.
  • Print Collections: State-specific practice guides and treatises in the library.

Institutions intending to adopt this assignment should ensure they have comparable coverage. At a minimum, they should have access to a full-text case database and secondary-source platforms, since the verification chain relies on the quality of materials available to students.

The Question Behind the Assignment

Ultimately, the project asks students to answer a deceptively simple question: Is negligence law truly uniform across the United States?

The research will likely suggest that the answer is both yes and no. The familiar framework of duty, breach, causation, and damages appears almost everywhere, but beneath that common structure lies a landscape of jurisdictional variation, doctrinal evolution, and competing policy approaches.

For first-year law students, uncovering that complexity bridges foundational research traditions with modern analytical tools. For law librarians and research instructors facing rapid technological change, this provides a practical guide for teaching essential legal research principles alongside AI literacy. These two areas should not be seen as competing methods; instead, they serve as complementary tools that enhance each other when used together effectively.