What do a women focused AI conference and an innocent comment on Reddit about public legal data have in common besides making me want to start burning things down? Well, stay tuned because that’s the point of today’s newsletter.
But first!
Let me give you an unsolicited and un-sponsored ad for Beehiiv, which is the platform you are reading me on right now. I am not a perfect consumer and I don’t think that it’s possible to be one. But folks…Substack is a Nazi bar. It’s been obvious for a while that they promote hate in the guise of “Free Speech” but this recent Guardian article really solidified it. I refuse to click on Substack links and will not promote works on it. I transferred over to Beehiiv last year. It wasn’t hard - literally push button transfer and if I wanted to monetize, I could. Please consider it. I’m happy to answer questions on how I did it.
BACK TO OUR PROGRAMMING.
I consider myself a feminist. I like women. But also…I like men. For most of my life, most of my friendships have been with males and my preferred social and professional activities are usually in very male dominated spaces. And honestly? I didn’t feel comfortable in or felt like I had many shared interests with other girls/women. I mean, this was me as a kid:
This isn’t me coming out as transgender, by the way. I have always felt my outside matched my inside. (But I am also supportive of other people’s journeys and feelings even if I can’t imagine what that feels like. Team Protect The Dolls 4eva.) My problem has always been that my insides didn’t match what SOCIETY wanted or expected from my outside.
Even when I joined a majority female profession, I somehow frequently ended up in spaces that were male dominated. I reacted in two ways: (1) going HAM on the Vera Bradley bags (2) wearing my lucky Appalachian Racoon Baculum necklace and if anyone asked about it say “Thanks, I thought I shouldn’t be the only person here without a penis and wanted to fit in.”
I’m delightful.
All of this is a round about way of saying…I don’t always love women focused events or feel the need to attend them. Especially when they throw pink over everything like it’s Shelby’s wedding and all the conversation is focused on how women are nurturing and empathetic and my neurodivergent ass is sitting there like “well fuck my drag.”
That’s right - call me Roy G. Biv, because I’m on the spectrum.
(I saw that on a t-shirt and was like “that’s amazing and I can’t wait to have an excuse to use it.” And here we are. You’re welcome.)
And it’s not that “soft skills” aren’t important or vital and traditionally women ARE better at it than men. Soft skills are actually very hard to do. But I’m in a real “Fuck you, Pay Me” kind of place mentally, emotionally, spiritually and I want to acknowledge that women can ALSO do the things that men can do as well as they can AND should be valued (for many meanings of the word) for that alone, not to mention the other assets they can bring.
I am pleased to report that the recent WAI conference managed to thread that needle of being empowering and supportive and highlighted the value women bring to the table when it comes to building and using AI without accidentally re-ghettoizing us as to be only the nurtures or supporters in an org. Women should be included in all aspects of AI development and use not just because we have - on average - special attributes that help improve the function of these tools, but also because we have the same abilities as the people traditionally called upon to do this.
And I didn’t see a goddamn pink balloon anywhere.
Me and some of my coworkers! And no goddamn pink balloons!
Cat kicked off the day with some slides showing the discrepancy of funding, design, use, etc of AI tools. Like, it is stunning how few women founded companies get funding on par to men founded one. And the people running the large AI companies? Gross. I mean, this event coming at the end of a week where it became clear how many leaders in technology, business, and academia regularly hung out with Jeffery Epstein was just….<chefs kiss>
(I can’t find original, but as someone on BlueSky said “it’s wild how many people were like “I think #MeToo has gone to far and I need to email Jeffery Epstein specifically about this. From my work email.”)
I wrote two notes to myself after this presentation. (1) I wish more men were in the room to hear this. Because just because an event is focused on women doesn’t mean that ONLY women should be there and right now we need more male allies to help change things (2) that being said, I think this was a great first step to have this message spread to so many women (men would have just taken up seats) because…it was affirming to realize “no, you’re not crazy, this is actually happening and it’s wrong.”
I think now, more than ever, it’s important to reaffirm that diversity is a strength. It makes better teams, it makes better products, and those two lead to more profits in long run. My colleague Stephanie gave a brilliant presentation on translating horror movie tropes to navigating the AI world. Someone pointed out that in horror movies, not only is the “final girl” white (and a “good girl” at that), but if you’re the Black friend in a horror movie? Oh, you’re dead in the first 30 minutes. Women have it hard in AI, and being a Black woman compounds that (by 10x at least.)
Which brings me to…
On my way back from WAI, this reddit thread was brought to my attention. The tl;dr version is OP asks [paraphrasing] “with rise of AI tools, do we still need the WEXISes of the world?” And there’s some good responses and bad responses and then someone innocently asked “Isn’t all the data in Lexis actually public data? What’s stopping people from just using publicly available data?”
I mean, would anyone blame me at this point?
Coincidentally, we are at just about the 10th anniversary of this study I did about the availability of legal information from government sources. I’m tempted to do an update but don’t think I have time, unfortunately. I’m a CC license gal so feel free to do it and remix, reuse, recycle my stuff if you have desire.
But just a couple of brief points…
Primary law is still extremely hard to get a hold of and use, for many reasons. (Read my report to start).
One massively positive change since I did that report is that the Free Law Project/Court Listener now can provide legal data . It’s so important to have a neutral third party that can be the source of this. It wasn’t that long ago that Fastcase couldn’t get a copy of the Georgia code from a rival.
That being said, primary law is more than case law. And a good AI training data set should have more than primary law. It needs a rich set of secondary materials to show the application of the primary law.
Right now, most AI is not only NOT trained on a decent legal corpus that has both a rich collection of primary and secondary legal materials, BUT it’s also trained on what we professional information scientists call “crap from the worst bowels of Internet Hell.”
Technology is not neutral. Technology will never be neutral because humans aren’t neutral. Technology development reflects both those who create the tools as well as the target audience (the audience that is prioritized) in the development process. So the goal is not to have a completely unbiased system, because that is impossible, but to have as many inputs and reviews as possible so that (1) biases can be discovered and then (2) mitigated as much as possible and accounted for.
The challenge now is to realize that human inputs matter as much as the content ones do as we navigate this new technology. But you can only do that when you notice or know which inputs are missing and make conscious effort to include them.
Be well,
Sarah
