A confident AI answer can cite a case that never existed. Teach people facing court to govern the machine, check the law, and build a research system they control.
Proposed for Chicago and Cook County: a legal-AI course for every learner, a hands-on research-harness lab for interested participants, a source-linked JUB tool, and new episodes that show the method in action. The foundation is Jim Winner's litigation-literacy framework and JUB's existing scripts, cited treatises, scenes, and assessments.
Access to a chatbot does not tell a tenant which deadline matters, whether a citation exists, or what to say when a hearing takes a turn.
of tenants observed at initial Cook County eviction hearings represented themselves.
Chicago Appleseed, 2025of tenants observed at Chicago initial eviction hearings had no attorney present.
Cook County court-watching report, p. 4of substantial civil legal problems reported by low-income Americans received no or insufficient legal help.
Legal Services Corporation, 2022percent of adults with a bachelor's degree versus a high-school education or less who had used ChatGPT in Pew's 2025 survey.
Pew Research Center, 2025The disparity is local, too. Black residents were less than 23% of Cook County's population but almost 51% of tenants observed in its eviction courts. Those are court-watching observations, not a countywide census. The AI-use figures are national and describe ChatGPT use, not legal competence. Together they show why a local program should teach both litigation and verification skills. Source
In February 2026, the National Center for State Courts reported more than 350 documented U.S. cases in which self-represented litigants cited nonexistent cases, statutes, or quotations generated by AI. Illinois courts have separately warned that fabricated authority in a filing can lead to sanctions. Teach citation checking before the first AI-assisted document reaches a clerk. NCSC ยท Illinois Courts
JUB has already taught the underlying skill. In the eviction course, all 36 learners with paired intake and final exams improved, and the average score rose from 66 to 83. Program viewing records show that approximately 47% of signups watched at least one instructional video. Jim Winner's litigation-literacy preprint
Start with the person using a chatbot tonight. Give interested learners the skills to build a repeatable research harness. Keep filming the legal moves both groups need to see.
Small cohorts read a real-looking notice, question an AI answer, inspect primary law, catch an invented citation, protect private facts, and rehearse a hearing. A pretest, scenario checks, and a post-test measure what they can do.
Proposed core courseParticipants assemble a local case workspace, source library, issue map, citation ledger, adverse-authority search, and review checklist. They test the system with planted errors, then keep the templates or application they built.
Proposed builder labNew episodes show how a made-up case, a stale rule, or a hidden assumption changes a real decision. Each starts with a cited treatise, passes through a writing room and legal review, and ends with a scene and a knowledge check.
Proposed new seasonJim Winner defines litigation literacy as the knowledge, skills, and dispositions needed to recognize a legal problem, use formal process, and participate competently with or without professional help. AI use now belongs inside that skill set.
Separate a termination notice from a summons. Find the case number, hearing date, requested relief, and source of each deadline before asking AI anything.
Define the jurisdiction, case stage, task, allowed sources, and stop condition. Use AI to explain unfamiliar terms and generate questions without sending private case papers to a public service by default.
Open the statute, rule, court form, or opinion yourself. Match the quoted words and holding to the original, check the date and court, then look for later or contrary authority.
Use a timeline, exhibit list, and claim-or-defense map. Mark what is known, missing, disputed, or needs professional review.
Practice a hearing, negotiation, or police encounter with branching scenarios. Try the next sentence when the other side interrupts, denies a fact, or offers a deal.
Challenge the answer with a planted false case, check every quotation, review privacy and local rules, and decide what needs counsel or more research before a person signs or files it.
An interested participant can leave with an enterprise-style research workflow for their own matter, assembled from free public law and ordinary tools. They see where each proposition came from and which questions remain open.
Save notices, orders, and evidence in a private case workspace. Record the source, date, and version. Use synthetic examples in class before anyone brings personal files.
Turn the facts into a dated timeline and an issue table. Each row states the question, governing forum, missing fact, and next research task.
Search official statutes, court rules, opinions, and forms first. Save a link and dated copy of every authority the AI proposes to use.
Resolve each case and quoted passage to the original. Record the court, date, procedural posture, relevant holding, and any later or contrary treatment found.
Feed the harness a nonexistent case, a real case with a false holding, an outdated rule, and a fact that flips the result. Fix the workflow when it lets one through.
Export a research packet with linked sources, open questions, and a human review log. The participant decides what to investigate further or bring to a lawyer.
Start with a source-backed case notebook in a spreadsheet or simple local app. Interested participants can then build an authority checker, an issue-to-JUB-scene navigator, or a branching hearing rehearsal. Each project stores the original source beside the proposed statement, marks missing support, and keeps an edit trail.
Start with public sources, not a paywalled AI answer. The lab will teach what each source covers, what it omits, and how to recheck a rule before use. An open source can support an enterprise-style process, but free databases do not guarantee complete docket coverage or a commercial citator's treatment analysis.
A spreadsheet, local files, and an optional no-cost or local AI model can run the first version. Model terms, data handling, limits, and fees must be checked before a participant connects any service. Federal docket records on PACER may still carry fees; the courts describe free access and exemption routes.
Participants need a tool that links JUB's video scenes and treatises to the original law, then gives them a way to challenge the answer. The Builder Lab teaches them to inspect and extend that same method.
Search a reviewed JUB issue index, jump to the exact video segment and treatise passage, open linked primary sources, build a timeline and issue map, and rehearse what changes when a fact changes. A citation ledger would show which propositions have a checked source and which still need research. Cohort results would be reported without publishing personal case stories.
NIST identifies confidently false content, privacy exposure, and over-reliance as generative-AI risks. These controls are the specification for a proposed product, not a claim of completed validation. Public release would require a current-law corpus, adversarial tests, privacy and accessibility review, and a clear boundary between legal information and individualized advice. NIST Generative AI Profile.
Viewers should not have to stop learning after the introduction. The proposal maps the next decisions, the hard turns, and the subjects they are already asking about.
Catch a fake case; check a real holding; build a source-linked research harness; govern the final AI-assisted work before a filing.
Discovery demands, hearing preparation, making a record, and the first appeal decision. Each episode branches when the opponent refuses, the judge rules, or a deadline changes.
Police-stop and school-encounter rehearsals that teach clear words, calm conduct, when to stay silent, and how to preserve the account for later review.
Debt collection and updated eviction problems chosen from resident questions and current local practice, with linked forms, sources, and what-if exercises.
This is an editorial proposal, not a claim that twelve episodes are already written, filmed, or approved. The existing library contains 26 finished episodes, with 19 on the public site across three series. The proposed season would use the established JUB sequence: cited treatise, writers' room, legal review, filming, edited episode, and learning check. Local rules and examples would be reviewed again before recording or release. See the existing-library scopes.
Equipment helps. The costly work would be teaching, producing, reviewing the law, testing safety, and getting the material to people before a deadline.
These are planning figures, not a Society offer or solicitation. The partnership brief lists separate archive and production examples, including a $1M concept with 20 original episodes and 60 short videos. The $1M AI concept below combines 12 episodes with teaching, a participant tool, and evaluation. A Society role would require a written license and separate approvals.
Build the six-session course and a small Builder Lab. Run Chicago and suburban Cook County pilots. Measure whether learners catch invented citations, identify the governing source, and keep private facts out of unsafe tools.
Release reviewed research-workspace templates and a limited source-linked JUB prototype. Train facilitators, produce the first AI-on-trial episodes, and test the tools against planted legal errors with residents.
Fund the twelve-episode season, a maintained participant tool, coached core and Builder Lab cohorts at multiple sites, language access, independent evaluation, and a year of source updates.
Report signups, unique viewers, one-lesson reach, cohort attendance, pretest and post-test scores, citation-check performance, and use of the issue index separately. Test whether participants can spot a fabricated authority, find adverse law, identify a deadline, and explain why a source does or does not govern. Count completed research harnesses only after a participant runs the error tests and documents a human review. Any later case-outcome study would require consent and a separate research design.
All scopes and allocations are planning figures, not vendor quotes or commitments. A final grant budget must match the selected entity, personnel plan, venue, and partner agreements.
This planning map considers instruction, production, equipment, and distribution. It does not assert that any named organization has agreed to participate.
Devices and creative instruction could support private research workspaces and media teaching. Apple's published community education work is a possible model, not a Society relationship or award.
Community Education InitiativeBuilder Lab workstations and local compute could help participants test tools without surrendering case materials. AMD's Learning Labs illustrate one possible model, not a Society relationship.
AMD Learning LabsMedia production and training could make source checks visible in a future episode. Sony's published guidelines are background research, not evidence of a Society request or relationship.
Sony giving guidelinesChicago residents building and testing their own AI research tools could fit a future education model. MacArthur's AI Opportunity program is not accepting unsolicited proposals. No Society proposal or award is represented here.
AI OpportunityLibraries, schools, community groups, or public agencies could be future distribution sites if an Illinois program is authorized. No Illinois site, buyer, grant, or pilot is committed.
Existing partnership briefExisting: 26 finished JUB episodes, including 19 on the public site, plus scripts, treatises, assessments, and an issue-based archive described in the partnership brief. Proposed: the six-session AI course, eight-week Builder Lab, participant research application, new season, sites, and annual scopes on this page.
This page shows an educational concept, not an active Society program or funding request. For current Just Us Bridge production questions, contact Professional Litigant, Inc. through the show website. A Society license, Illinois activity, and any funding would require separate decisions and compliance first.
Visit the Just Us Bridge ShowReturn to the Just Us Bridge partnership briefPublic legal education. Not legal advice. Proposed initiatives are not deployed services or funding commitments.