Artificial Intelligence: A Relentless March
In 2019, I wrote an article about the future role of artificial intelligence in writing legal briefs (see my previous post if you prefer a summary). Among other things, I observed that briefs usually follow a logical structure that computers can mimic. I speculated that AI might already be capable of generating basic briefs addressing simple issues—i.e., quashing subpoenas that violate express procedural rules or dismissing time-barred claims. But at the time, I lacked concrete examples.
That all changed an hour ago.
Earlier today, I asked OpenAI’s newly released ChatGPT application to write a simple motion. My prompt:
Write a motion to quash a subpoena in the U.S. District Court for the Southern District of Texas based on a violation of Federal Rule of Civil Procedure 45(d)(3)(a)(ii).
The bot's motion appears below. It’s not what I’d call a finished product, but it’s much better than I expected—certainly better than a lot of first drafts I’ve seen from junior lawyers.
Here it is in its entirety:
Motion to Quash Subpoena
Pursuant to Federal Rule of Civil Procedure 45(d)(3)(A)(ii), the undersigned moves to quash the subpoena issued by the plaintiff in this action, as it calls for the production of documents that are located more than 100 miles from the place of trial or hearing.
The documents in question are located at the offices of the undersigned's business, which is more than 100 miles from the Southern District of Texas. The burden and expense of producing these documents would be undue and unreasonable, given their distance from the place of trial or hearing.
Furthermore, the plaintiff has not shown any particularized need or relevance for the documents in question. The production of these documents would simply serve to harass and burden the undersigned, and should therefore be quashed.
Accordingly, the undersigned respectfully requests that the Court grant this motion to quash the subpoena and protect the undersigned from undue burden and expense. Respectfully submitted,
Not a final draft by any means; it’s missing case-specific details, and I object to a few of the bot’s style choices. Still, it’s pretty good: (1) it reads like a literate human wrote it, (2) it has a logical structure, and (3) it's refreshingly pithy.
This is just the beginning. Although some say that AI will eventually hit theoretical limits, those limits have yet to be discovered. Things just keep getting better. Be on the lookout for more.