Horseshoes, Hand Grenades, and Generative Text?

Hello Friends in the Computer (and Computer Friends if an AI reading this has become sentient),

The AI release waves are coming faster than the Covid variant ones. In attempting to clarify my thoughts and feelings, I fell back to my former webcomic days and I was inspired to make this.

I think encapsulates my feelings towards <waves hands in general direction of last few news cycles>.

I didn’t try ChatGPT-3 and have some misgivings about working with OpenAI products, but in a moment of weakness I finally got around to playing around with The Quattro. Like all self-absorbed bloggers, I immediately asked it about myself.

Not 100% factual but close. I have a very unique name and I am sure that helped. At least it didn’t identify me as a British law librarian like a lot of content farms have done ever since I worked at the University of Kentucky.

I did notice the site had a bunch of warnings, including this:

I don’t know what it says about my personality that I don’t worry so much about changing or breaking rules, but see how close to the line I can get and still achieve my goals. Is it laziness? I think it’s laziness.

So before we even get to legal advice or UPL (but oh honey we’re gonna talk about that one in a second), one issue people have that contributes to the access to justice crisis is knowing if they even have a legal problem. I do wonder if this technology (trained on proper data) could be used for something like that. So I did a quick and dirty test:

Enough there to feel like there’s some help being given without actually providing much in the way of insight, novelty, or assistance. Reminds me of most of the stuff I see on LinkedIn.

Okay, so let’s talk about The Quattro and A2J.

I realized that I should have added another layer to my chart between the purple and grey that said something like “annoyed at how this technology will be misused in the name of Access to Justice.

I’m not sure of this exact right term for this phenomenon for when any new advance in legal tech is announced. It’s close to greenwashing but not quite right. I call it “A2JazzHands.”

Usage: ”Sure this expensive, untested, and potentially insecure tool doesn’t have any proven use case yet, but think how it can help the access to justice crisis!” <jazz hands> 🤗

Today on the Legal Tech Week podcast I heard a common refrain that we need to stop saying, thinking, enabling, etc. A panelist was talking about using generative text tools for A2J and dropped the line “it’s not perfect but no help is better than none.”

No.

No.

No.

I want to make a few points abundantly clear:

1. People from impoverished, underprivileged, and marginalized communities should not have to accept “less than” just because it may be a free service, especially in legal or justice contexts.

2. Getting it wrong in a legal setting may actually be more actively harmful than doing nothing at all. See this thread from (former SCOTUS clerk and law professor now novelist) Courtney Milan for one example of Half-Ass2J can potentially really hurt people.

3. This statement, when used in the context of justice tech, seems to be assuming that the best possible solution for a problem is in person assistance from a barred lawyer and any tech intervention is automatically going to be less. I whole heartedly dispute that. The whole drive towards lawyer re-regulation is that so many access to justice scenarios DO NOT need the expertise of a human lawyer but current UPL rules make it too risky to do otherwise.

If I may quote my favorite Drag Queen Katya, “like, it’s so exhausting…having to confront everybody’s deficiencies day to day, so I’d like it if everybody were able to rise up a little bit so I didn’t have to be so tired.”

Let’s do better.

Be well.

Sarah

Thank you for reading Sarah's Commonplace Book. This post is public so feel free to share it.