Artificial intelligence has officially entered the legal profession. At this point, you've probably heard enough about its potential to last a lifetime, especially given that we've published blog after blog touting its usefulness in the legal profession.
It can draft, it can summarize, it can research, it can analyze, it can automate. The possibilities seem to be adding up all the time. And depending on who you ask, AI is either about to revolutionize the practice of law or bring civilization as we know it to a screeching halt.
For attorneys who haven't fully embraced the technology yet, all that noise can make AI seem more complicated than it needs to be, but you don't need a computer science degree to use AI effectively. You do need enough understanding to recognize what these tools can do, where they can go wrong, and when your professional responsibilities require you to slow down and think twice. That's exactly the balance explored in PBI's upcoming "Practical AI for Lawyers: The Tools and the Rules," a hands-on program designed to help attorneys understand both the practical applications of AI and the legal and ethical considerations that come with them.
Now let's take a look at a few simple questions surrounding AI that matter most to your practice.
What can this tool do? AI becomes much less intimidating once you stop thinking of it as one enormous, futuristic technology and start thinking about individual tasks. Generative AI tools can help brainstorm ideas, summarize lengthy information, create first drafts, compare documents, develop checklists, refine writing, and generate questions you may want to explore. Other AI-powered tools are increasingly being incorporated into legal research platforms, document review systems, contract management software, and programs attorneys already use every day.
Don't let that overwhelm you! Start small. Choose a low-risk, time-consuming task and experiment. Describe your workflow to AI and ask how it can help streamline some of your more menial tasks. It might surprise you.
What can't it do? This may be the more important question. Generative AI doesn't "know" information in the same way a human does. It generates responses based on patterns in the information and language on which its models operate. This is why they're called LLMs (Large Language Models). That means an answer can sound polished, confident, and completely plausible while still being totally inaccurate.
Lawyers have already learned this lesson the hard way through cases involving nonexistent judicial opinions and fabricated citations generated by AI. These are often referred to as "hallucinations." But hallucinations aren't limited to case law. AI can misunderstand facts, omit important context, oversimplify complicated issues, and make assumptions you never asked it to make.
Knowing a tool's limitations isn't a reason to avoid it. It's part of knowing how to use it.
Can I trust AI's output? Not automatically—and this is where human intelligence remains indispensable. AI can draft an argument, but the attorney must decide whether it's a good one. AI can summarize a case, but the attorney must confirm that the summary accurately reflects the opinion. AI can identify potential issues, but the attorney must determine which ones matter. AI can suggest language for a client communication, but the attorney still understands the client, the circumstances, and the consequences in a way the technology does not.
AI is very good at sounding authoritative, but you should treat it as an assistant more than anything. Verify citations. Review source materials. Check important facts. Read the final product as carefully as you would work prepared by anyone else. The faster AI makes it possible to produce information, the more important human judgment becomes in deciding whether that information deserves to be used.
What information am I giving it? Before entering information into an AI platform, consider what you're sharing. A prompt may contain client names, confidential facts, privileged communications, proprietary business information, or other sensitive material. Different AI products also have different policies governing how submitted information is stored, processed, and potentially used.
That makes understanding the tool itself important. Before incorporating an AI platform into legal work, attorneys should know what happens to the information they provide and whether their firm or organization has established policies governing its use. A useful rule of thumb is to pause before you paste. Convenience shouldn't override confidentiality.
How do I construct a successful prompt? A clearly stated prompt is what's going to determine the effectiveness of the AI at your disposal. You don't need to become a professional prompt engineer, but learning how to communicate effectively with AI will make a noticeable difference in its output.
"Summarize this" may produce something useful. "Summarize this document for an attorney preparing for a client meeting, identify the five most important issues, and flag anything that requires further investigation" gives the tool far more direction. Think of prompting less like entering keywords into a search engine and more like delegating a task. Provide context, explain the objective, and don't hesitate to give details. You're also free to specify the format you want and refine the results.
And remember: a better prompt can produce a better response, but it can't guarantee a correct one.
Is this actually making my work better? Just because you can use AI for something doesn't mean you should. If it takes fifteen minutes to construct a prompt, ten minutes to correct the response, and another fifteen to verify everything the AI produced for a task you could have completed yourself in twenty minutes, congratulations—you've successfully used cutting-edge technology to double your workload.
The goal isn't to incorporate AI into every task. It's to identify where it genuinely saves time, improves organization, sparks ideas, or removes repetitive work from your day. As you experiment, pay attention to where AI consistently helps and where it creates more work than it eliminates. Your most useful AI workflow probably won't look exactly like another attorney's and that's okay.
What rules apply? This is where "learning AI" becomes particularly important for lawyers. Using generative AI doesn't eliminate the professional obligations that already govern your practice. Attorneys still need to consider duties involving competence, confidentiality, communication, and other ethical issues. Courts, clients, employers, and individual organizations may also have their own requirements surrounding AI use.
That's why becoming comfortable with AI requires more than learning which button to click or how to write a clever prompt. Attorneys need to understand both the capabilities of the tools and the guardrails surrounding them.
Artificial intelligence is going to continue changing, and no lawyer will ever reach a point where they've learned everything there is to know about it. Six months from now, we'll probably be discussing tools and capabilities that aren't even on our radar today. The speed at which this technology is evolving is mind-blowing.
Fortunately, you don't need to know everything. Start with what the technology can and can't do for you. Verify what it produces. Experiment with the tasks where it can genuinely make your work easier. And above all, don't outsource the one thing your clients are actually hiring you to provide: your judgment.
