Vibe Coding For Lawyers: Next Big Thing Or Just A Big Dud?

There’s been lots of talk these days about vibe coding and whether it will be the next big thing in legal. It sounds so promising: lawyers, who spend their days billing six-minute increments while avoiding anything that resembles software development or understanding, can suddenly become coders. What could possibly go wrong?

So I was interested in a panel discussion at AALL this week entitled “Catch the Vibe: What Law Librarians Need to Know About Vibe Coding.” I hoped it would answer what lawyers need to know as well. The session was packed, suggesting that a lot of attendees had the same question as I did.

I need to point out that the panelists were all AALL members and serious tech people, pretty far up the curve with AI. So their take on whether vibe coding could be a practical tool was entitled to some weight. And while I’m no expert on technology and know nothing about coding, I am smarter than the average bear when it comes to tech.

Spoiler alert: I’m not sure vibe coding is ready for legal prime time, at least not yet.

What Is It?

The panel started with an explanation of vibe coding for those who have not been reading legal tech headlines. Vibe coding is what you can do when you find a solution to a problem that requires coding you have no clue how to write. So you ask the chatbot to write it for you. 

That right away gives some pause: you’re asking a chatbot to do something you can’t do, know little about, and aren’t able to evaluate when done.

The Demos: The First Hint

The panel started the discussion by talking about a vibe coding project each tried to do for the presentation. One tried to create an app that would synthesize data from various sources. A second panelist tried to create an app that would help teach concepts of Boolean search. And the third tried to use vibe coding to scrap information from various sources. 

Of the three, only the second panelist was marginally successful. All three spent substantial time trying. So while all were enthusiastic about the use, the fact that they had so much trouble was the first hint of problems.

The Concerns

The panel then moved to the risks and concerns with vibe coding. These, too, were telling: none of the panelists would dare use vibe coding on their employers’ systems. 

First, the panelists mentioned that while their work policies didn’t yet preclude vibe coding efforts directly, the policies were clear enough that none would risk it. (The projects each did were on their own computers on their own time.) 

Another telling sign: all were adamant that work policies should directly preclude vibe coding efforts. Why? In the hands of those without substantial expertise and sophistication, system security could be compromised. And the very architecture of the system could be threatened. As in all the files might disappear. 

Also, most workplaces have restrictions on where institutional data can be stored and used. Vibe coding by the unsophisticated could breach those policies and threaten the data. And it’s easy to overengineer a vibe coded app that could create unexpected issues and problems.

Another fear: people may try to vibe code without understanding what it is doing. This is not like asking ChatGPT for restaurant recommendations. The whole idea of vibe coding is that you create systems that can and will take actions without direct human involvement or approval. Great. Let it run and threaten both security and operations unbeknownst to anyone. 

And of course, like all things AI, it can hallucinate and create libraries that enable hackers to gain access. Bottom line, said the panelists, don’t use it for anything related to work.

And finally, it could take lots of tokens. So if you are a big user of AI for other things, its use for those other things might become constrained by playing around with something. Nothing like getting a big and unexpected bill to dim enthusiasm. 

How can creating computer code that you don’t understand and which could cause harm be a good thing?

So What’s It Good For?

The panelists were then asked what vibe coding might be good for given the concerns. A long silence …

Eventually, one finally said it can plan a party. One said maybe an interactive exercise. One said it’s good for creating games for their kids. The panelist that had some success with his demo said pulling data from various sources. Which, of course, is what his app sort of did. That’s about it. I wouldn’t call those ringing endorsements.

And all three were quick to point out that if you went beyond the very simple, you would need to get help from your IT department, the experts. Since lawyers in larger firms have these experts in house, going to them in the first place might save a lot of time and trouble.

But I Want To Vibe Code

Okay, I’m one of those who want to try it. What advice did the panel have? First, read up on it. See if someone else has already built what you are trying to. Duh. 

Don’t use it for something that would require ongoing maintenance and updating. Leave that for the experts. Use it for very simple projects where little can go wrong (which of course begs the question how would the unsophisticated know whether only a little could go wrong).

Another tip: start with the problem. Think it through carefully. Describe fully and in detail what you want the code to do and not do. Get fully educated on vibe coding before you let it rip.

Sounds like a lot of time of energy for little gain.

The Next Big Thing?

I walked into the panel discussion wondering whether next year’s conference might be all about vibe coding and the changes it will  wrought. 

After listening to these guys, I’m not sure. The hurdles seem pretty large. Just the time needed to figure out how to do it and create something useful seem significant for a business based on the billable hour. Vibe coding may be fun, but I doubt you can bill for it. 

And the risks if you don’t know what you are doing are large. Not to mention the fact that institutional policies don’t allow it. And IT departments shudder at the thought of cowboy lawyers vibe coding and threatening what they have to keep running. Most of the work to create a robust vibe-coded tool needs to be left to IT anyway, so why waste time trying to do it yourself. So, the work-use case for vibe coding right now seems pretty low.

Having said that, there could be a use case for those who want to tinker with it for creating simple personal uses like planning your next big party. There is also a place for sophisticated, and I underline sophisticated, solo or small firm lawyers to use vibe coding for work purposes, provided they know the risks. 

There may be instances where some have flirted with vibe coding without knowing it. Here’s a good example. I ask ChatGPT to search every morning for legal tech news, emphasize the U.S. market, rank the stories by importance, and deliver the report to me every day by 8 a.m. 

ChatGPT tells me that’s not vibe coding since I haven’t asked for any code to be created. But if I were to add to the list things like store the articles in a database or add a button that exports the report to Word, now I have asked the LLM to create a code and voila, I’m a vibe coder. Yippee.

And The Point Is?

The point is vibe coding may be similar to the development of AI use. We started with fear and loathing. Then we moved to “this is kind of cool for personal stuff like vacations and football summaries but not work.” Then it was “well, we can use it for work-related stuff, but we need prompt engineers.” To, “what the hell, I use it every day and it’s great. In fact, I use it and often don’t even know I am doing so, much less think about it.”

So we may see the same here. One thing I have learned with AI, what’s impossible today may become routine tomorrow. 

But right now? It’s not ready for prime time. Yet.

---

Stephen Embry is a lawyer, speaker, blogger, and writer. He publishes TechLaw Crossroads, a blog devoted to the examination of the tension between technology, the law, and the practice of law.