The duty of competence in the age of AI

Using AI in legal work is not just a practical skill — it's tied to an ethical duty of competence. Understanding this reframes AI from an optional efficiency tool into something lawyers are professionally obligated to handle correctly. This lesson previews the ethics rules (Module 3) by establishing the foundational duty.

Technology competence is an established duty. The ABA Model Rules' competence rule (Rule 1.1) includes, in Comment 8, a duty to keep abreast of "the benefits and risks associated with relevant technology" — adopted in the large majority of states. This "duty of technology competence" predates generative AI, but it now clearly extends to it: a lawyer using AI must understand enough about how it works to use it responsibly.

What competence with AI specifically requires (the ABA's first AI ethics opinion, Formal Opinion 512, spells this out — Module 3): a lawyer needs a reasonable understanding of the capabilities and limitations of the specific AI tool they use. You don't have to become a machine-learning expert, but you must grasp the essentials:

• AI hallucinates — it fabricates citations and facts (as Mata showed). Competence means knowing this and verifying.
• AI is not grounded in truth unless specifically designed to be — a general chatbot predicts plausible text, not verified law.
• AI has confidentiality implications — where does your input go, and is it protected (Module 3)?
• AI output must be reviewed — it is a starting point requiring professional judgment, never a finished product.
• Different tools have different reliability — a grounded legal-research platform differs from a consumer chatbot.

**Why framing this as a duty matters:**

• It's not optional to understand your tools. A lawyer who uses AI without understanding its fabrication risk and blindly files the output has failed the competence duty — that's the ethical framing of the Mata failure.
• It raises the bar on verification. Because competence requires understanding that AI hallucinates, filing unverified AI output isn't just careless — it's an ethics problem.
• It applies to supervision too — managing lawyers must ensure those they supervise (and vendors) use AI competently (Module 3's Rules 5.1/5.3).

The balanced view: competence cuts both ways. It doesn't require avoiding AI — used well, AI is a legitimate, valuable tool, and refusing to learn it may itself become a competence gap over time. What competence requires is understanding the tools well enough to use them responsibly: knowing their limits, verifying their output, protecting client data, and applying professional judgment. Competence is the bridge between "AI is useful" and "AI is used ethically."

The mindset: using AI competently is an ethical duty, not merely good practice — the duty of technology competence (Rule 1.1, Comment 8) extends to understanding AI's capabilities and, crucially, its limitations. That means knowing AI hallucinates, that it's not inherently grounded in truth, that it has confidentiality implications, and that its output requires review and verification. This duty is why the Mata failure was an ethics problem, not just a mistake, and it's the foundation for the rules and workflow the rest of the course builds. Understand your tools well enough to use them responsibly — that's competence in the age of AI.