Why access fails

01

An anonymized lived account

$364 paid for one hour. The participant recalls spending about forty-five minutes telling the story again.

Emails from February 2026 record a one-hour private consultation at CAD $364 including tax. The firm required the documents and payment by 3 p.m. so the lawyer could review them before the 4 p.m. appointment; a later invoice email confirmed that the service had been paid. The participant recalls that about forty-five minutes were nevertheless spent reconstructing the events and that the documents were not meaningfully used. That account of the call is participant testimony, not a recording.

The consultation did not create an ongoing route to help. Five months later, the participant sent the same firm a new written chronology and asked for the documents to be reviewed in advance. The firm replied that it was unable to assist. Other free and low-cost routes led to short appointments, unavailable appointments, limited document review or another referral back to the private market.

During a short human-rights clinic appointment, the participant recalls being told that the situation was not viewed as a human-rights matter. No recording or written advice confirms those words, so the public record labels them as recollection. Later, source-linked research identified facts that could engage a protected ground and a possible BC Human Rights Tribunal route. The participant used a Tribunal pathway; filing does not establish the merits of the underlying complaint.

02

What the contemporaneous emails show

Seven dated records trace one person through the same referral loop.

• February 20: the private firm quoted CAD $364 including tax for one hour and required payment and documents in advance so the lawyer could review them before the appointment.
• February 24: the firm confirmed that the consultation had been paid.
• June 15: a pro bono service wrote that it had no appointment available in the relevant area of law and invited the participant to check again in two weeks.
• July 16: a referral service described a free 15–30 minute introduction that assesses whether an issue is worth pursuing, but does not include legal advice or services; any further retainer is fee-paying.
• July 20: the human-rights clinic wrote that it had no time to review the 59-page attachment. Its appointment terms allowed up to 30 minutes and did not guarantee further help or representation.
• July 24: an advocacy program wrote that the matter had “some complicated and nuanced issues that would be better assisted elsewhere,” suggested paid private counsel and recommended a full-service employment lawyer.
• July 24: after the participant again supplied a detailed chronology and offered the documents in advance, the original private firm wrote that it was unable to assist.

03

The wider pattern

Millions of people meet legal problems without sustained legal help.

Statistics Canada’s 2021 survey found that 18% of adults in the ten provinces—about 5.5 million people—experienced a serious problem or dispute that was not easy to fix. Only 33% contacted a legal professional. Fifty-one per cent searched the internet.

Among people who took action but did not contact a lawyer, 37% said they could not afford legal help. Fifty-two per cent did not initially understand the possible legal implications. These figures do not prove that every person needed a lawyer. They show why a system that depends on early issue recognition and self-directed research will predictably lose people.

SourcesStatistics Canada

04

Service design

A short consultation is useful. It is not representation or case preparation.

Access Pro Bono describes a free 15–30 minute Legal Referral Service consultation and up to half an hour through its Summary Advice Program. The BC Human Rights Clinic says demand exceeds resources, service is not guaranteed and short-service appointments last up to 30 minutes and book quickly.

Those services have finite mandates. The gap appears when the task is larger than the appointment: reading a record, testing more than one legal characterization, comparing decisions, building a chronology and returning when new evidence changes the picture.

SourcesAccess Pro Bono · Access Pro Bono · BC Human Rights Clinic · BC Human Rights Clinic

05

The research proposition

This is where AI can be the only research assistance left.

Scott Shapiro described Yale’s target as “Cases in which it’s either AI or nothing.” The lesson is not that a chatbot should replace a lawyer. Yale explores formal reasoning, narrow tasks and distribution through legal-aid and pro bono organizations.

General-purpose AI is not a safe substitute for source work. A preregistered Stanford evaluation found hallucinations above 17% in the tested LexisNexis and Thomson Reuters systems. A separate US federal-case benchmark found higher rates for general-purpose models. Those are results for specific tests, not a universal error rate, but they rule out uncited answers as a public-service standard.

SourcesYale News · Stanford Law School · arXiv

06

Design consequence

The tool must show its work.

• Separate the participant’s facts, the system’s inference and the source text.
• Link each legal proposition to an original decision, statute, rule or official procedural page.
• Show the passage and pinpoint, why it matched, and differences that may make it inapplicable.
• List missing dates, facts and documents before offering a roadmap.
• Preserve the query, filters, corpus version and result set.
• Export a pack a lawyer, clinic or institution can review without making the participant start again.