Module 1 of 11

AI Literacy & the Paralegal Role

Understanding how AI works, where it fails, and what your professional obligations require before it touches a client's case.

Every Tool Comes With an Obligation

A good paralegal in a criminal defense practice knows what each database searches, what each form requires, what each filing deadline means. The same standard applies to AI. Before using these tools on anything that touches a client's case, you need to understand how they work, where they fail, and what your professional obligations require when you use them.

Working within a supervised legal practice, in a field where errors have direct consequences for real people facing potential loss of liberty, sets a higher standard than casual AI use. Build the right habits from the beginning.

Learning Objectives

Explain how large language models generate text and why this produces hallucinations.
Identify the three failure modes most dangerous in criminal defense paralegal work.
Apply the verification protocol to any AI output before it enters a case file.
Locate your jurisdiction's paralegal ethics rules and identify AI-specific guidance.
Build a personal AI use framework grounded in supervision and verification.

Chapter 1 Reading Notes

📚 The AI-Powered Criminal Defense Paralegal · Chapter 1

How LLMs Work — and Why It Matters

Large language models predict the next most probable token in a sequence. They are not retrieving facts from a database. They are not reasoning through a problem. They are generating plausible-sounding text based on patterns in training data. The result: AI can produce confident, well-formatted output that is entirely fabricated — and that output looks identical to accurate output.

"The paralegal who understands this builds verification into every workflow. The paralegal who does not will eventually submit a hallucinated case citation to their supervising attorney."

Three Failure Modes That Matter in Criminal Defense

Citation Hallucination. AI cites real-sounding cases that do not exist, or real cases that do not hold what AI claims. Every citation must be verified in Westlaw or Lexis before it enters any work product.

Jurisdictional Error. AI trained on national data may state law from the wrong jurisdiction. A suppression standard from California is not the standard in North Carolina. Always verify jurisdiction-specific rules.

Confidentiality Risk. Entering client facts into a free consumer AI tool is a potential unauthorized disclosure. Know which tools are covered by data protection agreements before entering any case information.

The Verification Protocol

For every AI output before it enters a case file: (1) Source check — does a primary source support this claim? (2) Citation check — does this case/statute exist and say what AI claims? (3) Jurisdiction check — does this rule apply in your jurisdiction? (4) Currency check — is this still good law / current rule? (5) Attorney review — has supervising counsel approved the final work product?

Activity 1 · AI or Attorney Judgment?

Click each task to see whether it is appropriate for AI assistance or requires attorney judgment.

Activity 2 · Verification Checklist

Run every AI output through this checklist before it enters a case file.

Activity 3 · Platform Quick Reference

Click each tool type to see what it is good for and what it cannot do in criminal defense work.

Practice Quiz · Module 1

10 questions — unlimited attempts, highest score saved.

Module 1 Assessment

12 questions · 2 attempts · 75% to pass.