Poke around this blog and site, and you’ll see why I talk about these things, what I think, what I do to help. You might also be curious about LawQi.

But enough about me. So let’s get into it.

Here are 30 reasons, grounded in common legal ethics concerns, court rules, and practical considerations, that I hear from lawyers about why they are hesitant to use AI (specifically Large Language Models like ChatGPT):

1. Hallucinations/Inaccuracy: LLMs can generate plausible-sounding but factually incorrect information.
2. Lack of Legal Reasoning: LLMs don’t “understand” the law; they predict text, not legal principles.
3. Confidentiality Breaches: Inputting client data into an LLM could violate attorney-client privilege and data privacy regulations (like GDPR, CCPA, PIPEDA, etc.).
4. Unauthorized Practice of Law (UPL): Relying solely on an LLM for legal advice could be considered UPL by a non-lawyer (the LLM).
5. Duty of Competence: Lawyers have a duty to provide competent representation, and over-reliance on an unproven technology could violate this duty.
6. Duty of Supervision: Lawyers must supervise non-lawyer staff (including technology) appropriately, which is difficult with complex LLMs.
7. Bias and Discrimination: LLMs can perpetuate biases present in their training data, leading to unfair or discriminatory outcomes.
8. Lack of Transparency/Explainability: It’s difficult to understand why an LLM produced a particular output, making it hard to verify its reasoning.
9. Data Security Risks: LLM providers could be targets for hacking, leading to data breaches of sensitive client information.
10. Unpredictability: LLM outputs can vary even with the same input, making them unreliable for consistent legal work.
11. Over-Reliance/Deskilling: Lawyers might become overly dependent on LLMs, diminishing their own legal skills.
12. Court Rules/Procedure: Courts may not accept LLM-generated documents or arguments without significant lawyer oversight and certification.
13. Malpractice Liability: Using an LLM that makes an error could increase a lawyer’s liability for malpractice.
14. Lack of Case Law Precedent: There’s limited case law on the use of LLMs in legal practice, creating uncertainty.
15. Ethical Guidelines are Unclear: Bar associations and regulators are still developing guidance on the ethical use of LLMs.
16. Client Consent Issues: Clients may not consent to their information being used by an LLM.
17. Cost vs. Benefit: The cost of LLM services might outweigh the benefits for some practices.
18. Integration Challenges: Integrating LLMs into existing legal workflows and software can be difficult.
19. Version Control/Updates: LLMs are constantly updated, making it hard to track the specific model used for a particular task.
20. Copyright Issues: LLM-generated text could potentially infringe on existing copyrights.
21. Evidence Admissibility: LLM-generated content might not be admissible as evidence in court.
22. “Black Box” Problem: The internal workings of LLMs are often opaque, making it hard to audit or debug them.
23. Lack of Human Judgment: LLMs lack the nuanced judgment and ethical considerations that human lawyers bring to legal practice.
24. Reputational Risk: Using an LLM that produces a poor result could damage a lawyer’s reputation.
25. Insurance Coverage Uncertainty: Malpractice insurance may not cover errors made by LLMs.
26. Difficulties in Prompt Engineering: Getting the desired output from an LLM requires skilled prompt engineering, which is a new skill for many lawyers.
27. Jurisdictional Issues: LLMs may be trained on data from one jurisdiction but used in another, leading to incorrect legal advice.
28. Lack of Personal Accountability: It’s difficult to assign responsibility for errors made by an LLM.
29. Training Data Limitations: LLMs are only as good as the data they are trained on, which may be incomplete or outdated.
30. Regulatory Uncertainty: Future regulations could restrict or prohibit the use of LLMs in legal practice

One reason they should start

The legal profession has always adapted to advancements in information technology, integrating new tools while maintaining its core ethical obligations. The rapid evolution of LLMs presents another such moment. Concerns about reliability, security, and ethical implications are precisely the issues that can only be answered through direct engagement.

Given the accelerating pace of LLM development and adoption across society, a passive stance is no longer tenable. Controlled experimentation, within the bounds of our existing professional standards, is not merely permissible; it is becoming essential to understand how these tools can, and must, be wielded responsibly to serve our clients’ best interests. The fundamental duty remains: competent, ethical, and informed representation. Mastering the evolving information landscape, including LLMs, is now an inseparable part of that duty.