Latest / AI Tools for Practicing Lawyers / Field Note: How BigLaw Associates Are Actually Using AI in Legal Drafting

Transcript

Field Note: How BigLaw Associates Are Actually Using AI in Legal Drafting · AI Tools for Practicing Lawyers

• 0:03
  SPEAKER_00: Welcome to AI Tools for Practicing Lawyers.
• 0:07
  SPEAKER_00: Practical, no-nonsense guidance to help attorneys put AI power to work in their practice right now.
• 0:16
  SPEAKER_01: This field note is a direct companion to the one I just released.
• 0:21
  SPEAKER_01: The one on 21 ways AI can hallucinate in your legal brief.
• 0:26
  SPEAKER_01: If that episode was about how AI fails, this one is about how lawyers are adapting anyway.
• 0:33
  SPEAKER_01: Because here's the reality: AI hallucinations are not going to stop lawyers from using these tools, but they are going to force lawyers to use them differently.
• 0:44
  SPEAKER_01: This is not a CLE presentation.
• 0:47
  SPEAKER_01: It's not a vendor demo.
• 0:49
  SPEAKER_01: It's not a law firm memo.
• 0:51
  SPEAKER_01: This comes from a real-world Reddit thread.
• 0:55
  SPEAKER_01: It's about 75 comments from big law associates in the big law subreddit talking about how they're actually using AI in drafting.
• 1:06
  SPEAKER_01: And what's really interesting is the lawyers aren't saying don't use AI at all.
• 1:14
  SPEAKER_01: What they're saying is, here's how I use it carefully, and it works.
• 1:20
  SPEAKER_01: And what emerged from that thread were patterns, not so much tips and tricks, although those you could say those are in there too, but patterns.
• 1:31
  SPEAKER_01: And once you see the patterns, you start to understand something important.
• 1:36
  SPEAKER_01: AI is not replacing legal drafting, it's reshaping how drafting gets done.
• 1:43
  SPEAKER_01: So here's the underlying principle before we get into specifics.
• 1:48
  SPEAKER_01: AI is being used for structure and volume, but not judgment.
• 1:54
  SPEAKER_01: The lawyers getting the most value are using AI to start faster.
• 1:59
  SPEAKER_01: That's the big thing.
• 2:01
  SPEAKER_01: They want to battle the blank page.
• 2:03
  SPEAKER_01: They organize better, they reduce mechanical effort.
• 2:08
  SPEAKER_01: They're not using it for legal authority or final conclusions or anything that requires trust.
• 2:15
  SPEAKER_01: That's the dividing line.
• 2:17
  SPEAKER_01: So let's walk through some of these most important patterns.
• 2:22
  SPEAKER_01: As I said before, number one, breaking the blank page.
• 2:27
  SPEAKER_01: The hardest part of drafting isn't writing, it's starting.
• 2:31
  SPEAKER_01: You know, I remember Mel Brooks talking to his wife Ann Bancroft and complaining about how horrible it is dealing with the blank page as a screenwriter.
• 2:43
  SPEAKER_01: The same thing is true for lawyers.
• 2:45
  SPEAKER_01: The hardest part is dealing with that blank page.
• 2:50
  SPEAKER_01: And most associates said the same thing that AI eliminates the blank page mind problem.
• 3:00
  SPEAKER_01: Because even if they throw away most of the output, which happens, having something on the page changed the task from I need to write this to I need to fix this.
• 3:12
  SPEAKER_01: And really, that's a completely different cognitive load.
• 3:17
  SPEAKER_01: But here's the catch AI output is wordy.
• 3:20
  SPEAKER_01: I mean, anybody who spends time brainstorming Chat GPT, they're going to know that this bot doesn't stop talking.
• 3:29
  SPEAKER_01: But it also hedges, it overexplains.
• 3:34
  SPEAKER_01: So you need to treat the work product that the AI has spit out to you as scaffolding.
• 3:41
  SPEAKER_01: It's not finished work.
• 3:43
  SPEAKER_01: All right, the second is a very popular strategy is to draft in small pieces.
• 3:50
  SPEAKER_01: So one of the most upvoted approaches was don't ask for a full brief.
• 3:55
  SPEAKER_01: Ask for one or two paragraphs at a time.
• 3:58
  SPEAKER_01: Feed in your cases and your quotes and your argument, then have AI draft just that section.
• 4:05
  SPEAKER_01: So this works because you stay in control.
• 4:09
  SPEAKER_01: And it matters because full document prompts create full document problems.
• 4:16
  SPEAKER_01: All right, so now another, let's say, let's call this number three, is structure before prose.
• 4:24
  SPEAKER_01: This one mirrors how good lawyers already think.
• 4:27
  SPEAKER_01: Ask for an outline first, then expand it.
• 4:30
  SPEAKER_01: Because here's the truth.
• 4:32
  SPEAKER_01: If the structure is wrong, the brief is wrong, no matter how polished it looks.
• 4:37
  SPEAKER_01: AI makes it easy to skip that step.
• 4:40
  SPEAKER_01: Big law associates are deliberately not skipping it.
• 4:44
  SPEAKER_01: So what they do is they create an outline.
• 4:48
  SPEAKER_01: Maybe they use AI to create the outline, and then they flesh out each item from the outline until they have a brief that they can then go back and revise.
• 5:02
  SPEAKER_01: So item number four, turning facts into arguments.
• 5:07
  SPEAKER_01: This is where AI starts to become genuinely powerful.
• 5:12
  SPEAKER_01: So the associates are feeding in raw facts, client notes, messy inputs.
• 5:19
  SPEAKER_01: And let's make sure that you're using an AI tool that's at least at the enterprise level if you're going to be putting in client notes, because that's really a privilege item.
• 5:32
  SPEAKER_01: It's very much a yellow light item.
• 5:36
  SPEAKER_01: So they they're asking AI to produce a structured argument framework, not just an outline, but a meaningful organization of the facts.
• 5:48
  SPEAKER_01: But there's a caution here, which is that AI organizes based on patterns it has learned, not necessarily based on your judge, your jurisdiction, or your strategy.
• 6:01
  SPEAKER_01: So you still have to think.
• 6:05
  SPEAKER_01: All right, number five, iteration, not one-shot prompting.
• 6:09
  SPEAKER_01: The lawyer is getting value from AI.
• 6:11
  SPEAKER_01: They're not just sticking a prompt, getting an output and saying, all right, I'm done.
• 6:16
  SPEAKER_01: They are prompting and then reviewing the output, adjusting, reprompting, then refining multiple passes, having a conversation with AI.
• 6:28
  SPEAKER_01: This is the way, frankly, this is the way a lot of AI addicts are working.
• 6:33
  SPEAKER_01: They're having deep, detailed conversations with AI, and sometimes jumping back and forth from one AI tool, let's say ChatGPT, for brainstorming, to another AI tool, let's say Claude for drafting.
• 6:49
  SPEAKER_01: And then you take what Claude drafted, you dump it back into ChatGPT, and you ask for ChatGPT's input.
• 6:56
  SPEAKER_01: This has become a very common workflow for people who are a little bit more on the Jetsons level.
• 7:03
  SPEAKER_01: But you know, still could be advanced Simpsons.
• 7:07
  SPEAKER_01: So anyway, one associate even noted that asking AI clarifying questions before drafting reduced correction time later.
• 7:17
  SPEAKER_01: I mean, that's a subtle, but still very important shift.
• 7:21
  SPEAKER_01: All right, number six, cleaning up messy drafts.
• 7:25
  SPEAKER_01: This is one of the safest and most effective uses.
• 7:30
  SPEAKER_01: Take rough writing and ask AI to improve flow, organize issues, clarify structure.
• 7:38
  SPEAKER_01: One example from the thread was turning a messy contract analysis into a clean issue table.
• 7:45
  SPEAKER_01: But here's the risk: AI can change meaning while improving clarity.
• 7:51
  SPEAKER_01: So you always compare the cleaned version to the original.
• 7:55
  SPEAKER_01: A lot of these associates like to talk about treating A like a junior associate's.
• 8:01
  SPEAKER_01: So this was the most upvoted comment in the thread.
• 8:04
  SPEAKER_01: I treat it like a first-year associate.
• 8:06
  SPEAKER_01: Now, what does that really imply?
• 8:08
  SPEAKER_01: You give instructions, you expect imperfections, you review everything, you take responsibility.
• 8:16
  SPEAKER_01: That framing was incredibly useful because it prevents two mistakes: overtrusting AI and underutilizing AI.
• 8:27
  SPEAKER_01: All right, let's go on to number eight.
• 8:30
  SPEAKER_01: Tone and clarity rewrites.
• 8:32
  SPEAKER_01: This is low risk, high return.
• 8:34
  SPEAKER_01: Associates are using AI to tighten language, adjust tone, simplify explanations, especially for things like client emails, explanatory sections, long communications.
• 8:48
  SPEAKER_01: I mean, one of my favorite things to use AI for is going through a massive long email chain and having AI just summarize what's really going on in this email chain.
• 9:03
  SPEAKER_01: Anyway, you'll notice that AI has a voice.
• 9:07
  SPEAKER_01: It tends to sound kind of formal, maybe cautious and hedged.
• 9:12
  SPEAKER_01: So you may need to put your voice back into the draft.
• 9:16
  SPEAKER_01: All right, another thing they like to talk about in the Reddit sub thread were chunking complex arguments.
• 9:24
  SPEAKER_01: So instead of drafting one long argument, they break it into one issue at a time, one section at a time.
• 9:32
  SPEAKER_01: This reduces errors and increases precision.
• 9:35
  SPEAKER_01: But of course, it creates a whole new task, which is stitching everything back together.
• 9:40
  SPEAKER_01: And that requires a full read-through for consistency.
• 9:46
  SPEAKER_01: All right, another item is using prior work product.
• 9:50
  SPEAKER_01: And this is where things get powerful and risky.
• 9:53
  SPEAKER_01: Associates are feeding AI prior briefs, memos, precedent language, and using that to guide the drafting.
• 10:01
  SPEAKER_01: The result is significantly better output, no question about it.
• 10:06
  SPEAKER_01: But it raises a serious issue.
• 10:08
  SPEAKER_01: I mean, there's work product governance.
• 10:11
  SPEAKER_01: You need to know where your data is going, whether it's being retained, whether it's being used for training.
• 10:17
  SPEAKER_01: This connects directly to the three-legged stool framework.
• 10:21
  SPEAKER_01: And I would ask you to go back and look at least at the show notes of episode 008 and the field note that was a companion to that, which was how to implement the three-legged stool for discovery compliance.
• 10:37
  SPEAKER_01: All right, number 11, AI is a thinking partner, not just drafting, but thinking.
• 10:43
  SPEAKER_01: Associates are asking, what are the weaknesses in my argument?
• 10:46
  SPEAKER_01: What counterarguments exist?
• 10:48
  SPEAKER_01: What am I missing?
• 10:49
  SPEAKER_01: This is really one of the most interesting use cases because it shifts AI from a writing tool to a thinking tool.
• 10:57
  SPEAKER_01: But again, AI's thinking is pattern-based.
• 11:01
  SPEAKER_01: It may miss judge-specific tendencies, local rules, strategic nuance.
• 11:08
  SPEAKER_01: All right, then the number 12 item was avoiding AI for citations.
• 11:13
  SPEAKER_01: This was the most consistent agreement in the entire thread.
• 11:17
  SPEAKER_01: Do not trust AI for case law.
• 11:21
  SPEAKER_01: Even lawyers who use AI heavily said, I just can't trust it with research.
• 11:25
  SPEAKER_01: And if you listen to the prior field note about 21 ways that AI can hallucinate in your legal brief, you'll understand exactly why.
• 11:34
  SPEAKER_01: This is the bright red light line.
• 11:37
  SPEAKER_01: The pattern underneath when you step back, here's what you see.
• 11:42
  SPEAKER_01: AI is being used where drafting is mechanical, repetitive, structurally heavy, and it's avoided where the drafting requires accuracy, judgment, and authority.
• 11:55
  SPEAKER_01: I mean, another way you could put it is that AI handles the work, but the lawyer handles the responsibility.
• 12:02
  SPEAKER_01: Now, here's where this gets real.
• 12:05
  SPEAKER_01: It gets very real.
• 12:07
  SPEAKER_01: Big law associates have something you may not have, and that's a safety net.
• 12:11
  SPEAKER_01: Their work is reviewed by senior associates, partners, multiple layers.
• 12:17
  SPEAKER_01: Listen, if you're in a solo or a small law firm, that safety net is smaller or non-existent.
• 12:24
  SPEAKER_01: So the workflows they become riskier.
• 12:28
  SPEAKER_01: They become more dependent on your discipline.
• 12:33
  SPEAKER_01: As a final thought, the question isn't should lawyers use AI in drafting?
• 12:39
  SPEAKER_01: I mean, that question's been answered.
• 12:40
  SPEAKER_01: They're doing it.
• 12:41
  SPEAKER_01: The lawyers are using AI in drafting.
• 12:44
  SPEAKER_01: But the real question is: do you know exactly where AI stops being reliable?
• 12:49
  SPEAKER_01: Because the lawyers who understand that line are going to move faster and safer than everyone else.
• 12:58
  SPEAKER_01: This field note was designed to work together with the 21 ways that AI can hallucinate in your legal brief.
• 13:06
  SPEAKER_01: The first one shows you how AI fails.
• 13:08
  SPEAKER_01: This one shows you how lawyers are adapting.
• 13:11
  SPEAKER_01: You should use both.
• 13:13
  SPEAKER_01: Because AI is not removing judgment from legal writing, it's increasing the importance of it.
• 13:19
  SPEAKER_01: This is Ron Drescher, and this is AI Tools for Practicing Lawyers.
• 13:24
  SPEAKER_01: And I'll see you in the next episode.
• 13:31
  SPEAKER_00: That's it for today's episode of AI Tools for Practicing Lawyers.
• 13:36
  SPEAKER_00: Thanks for listening.
• 13:37
  SPEAKER_00: We'll see you next time.