How Soon is Now?

Pharmacist and I went to The Legal Forecast launch event. Before I start, I have to say, the food they had at this event, was far fancier than any other event I've been to, including conferences and corporate events (for those wondering, that ring-shaped brown thing was bread).

What I gathered from the launch is that TLF is about promoting innovation in the legal field, which, as someone who works in IT, I'm all on board for. They didn't really go into any details of current innovations in particular, but the keynote speaker discussed how important innovation is, and you don't want to be left behind. Also on board that train.

However, innovation is a double-edged sword. I met up with a friend who is currently working as a pick-packer. She works in a warehouse, and gets a list of orders for things which she needs to grab from the shelf and put into boxes which get mailed out. Sound like something that Amazon has an army of robots to take care of?

Amazon bought a robotics company called Kiva Systems in 2012 for $775 million (£632 million). Kiva’s robots automate the picking and packing process at large warehouses in a way that stands to help Amazon become more efficient. The robots — 16 inches tall and almost 145kg — can run at 5mph and haul packages weighing up to 317kg.

Human pick-packing is a dying industry, and my friend complained about how her job will soon be obsolete. How can you compete with a worker who never complains, never gets tired or sick, and doesn't feel the need to form a union and demand better working conditions (at least not yet)?

A similar thing is happening within the legal industry, with what Pharmacist calls "robo-lawyers" becoming more and more popular, as people write apps that help other people deal with simple legal tasks, like writing a letter to contest a fine, or helping people work out whether they actually need legal advice or not. It could be argued that these apps free up time of lawyers so they can work on cases that do require human processing, and I think with sites like https://finefixer.org.au/ (seems to be Victoria-based, with legal advice from the Moonee Valley Legal Service), it also means that people who were otherwise unaware that they had other legal options can work out what to do.

What Pharmacist and I discussed at the end of the night was how advanced these algorithms could become (one of the speakers dropped the A-word, and I got kinda excited, but she didn't go into it in depth). In the case of things like fines, it's pretty much just directing someone through a flow chart, trying to work out the circumstances of the fine, and whether there would be room for contesting it. You have a set list of criteria, and you gather the facts of the situation, and see whether those criteria match.

I mean, that's what law boils down to, if you think about it. There are all the laws, there are the past cases that have been tried, along with their outcomes, and then there are all the situations that these need to be applied to. The job of a lawyer is to know how to apply the former to the latter. And if there's anything a computer is good at, it's following rules.

The only problem is that there is currently no "language" for these rules. Similar to how one person might go to the doctor and say, "I feel an acute pain in the first metatarsal of my left foot", and another person might say, "My foot hurts", with both of them are referring to the same type of injury, different people will describe their legal problem in different ways, and we don't currently have a way to parse these problems into something we can use to search a legal database with. As Pharmacist said to me, one of the biggest issues is parsing all that knowledge into something searchable. But it really is just a matter of time.

If you think about it, it wasn't too long ago that most people had no way to interact with computers. We evolved from punch cards to the first word-based programming languages (that were incredibly low-level), to higher level programming-languages, which still required technical knowledge. If you wanted to calculate something, you either had to learn to code, or have someone else do it for you.

Then we developed operating systems, and moved from keyboard/moused-based interaction to touch-screen interaction, most of which can be done by toddlers. But the task of creating and maintaining apps is still a specialised skill.

I wonder if lawyers will one-day evolve to something like that. They become the interface between people and the legal databases: translating human descriptions of problems into a language that can be processed by software used to search past cases. Court cases will be entered into the database as they happen, ensuring all the rulings stay current.

The question is: why can't that translation step be automated? At the end of the day, someone is going to be teaching these "lawyers" how to translate layman speech into legalese, and if there are classes on this, then it's likely that eventually, this knowledge can also be boiled down to a series of rules - similar to how Google seems to just "know" that when I search for a "string", I'm talking about a series of characters in programming, and not the thing you used to tie to a kite. (One of my lecturers did a demo in class of him searching for something that sounds "normal", but is actually the name for something "techy", and showed that the techy result came up for him first in Google, but for his friend, the non-techy thing came up first. I thought the term was "string", but a techy result is coming up first for both me and Pharmacist (who isn't a programmer). Oh well, you get the gist).

I have a feeling that while most of us can sleep soundly with the idea that there are robots crawling around grabbing items from shelves and putting them into boxes, the thought that the fate of your life lies in the decision of a robot wouldn't sit so comfortably. I mean, as a programmer, I would be wondering how rigorous the testing for the Judge 2.0 software was, or whether someone had slipped in a backdoor to allow them to rig a decision of they really needed to. A lot of people don't even trust voting machines! I imagine such software would have to have an audit trail the size of the solar system, but the average person won't trust it. Same with having a lawyer. If I were getting divorced, I wouldn't want two computer programs fighting over who gets what, or even one computer program taking a series of inputs from me and my soon-to-be-ex husband and just printing a list of the results. Even though I feel that this would be the most logical way to sort it out, it just feels so cold and impersonal.

Which is why I personally don't see the legal system being completely replaced by an automated system anytime soon. I definitely think aspects of it can be automated, and given how my lawyer friends always complain about having a never-ending backlog of cases to look at, I don't see there being a backlash at automating simple tasks. The engineer side of me would love to be proved wrong, and the idea of a truly unbiased judge (if that's even possible, as the software can also be biased) also appeals to the philosophical side of me, but I feel like the psychological side of me thinks that humans are always going to prefer the human touch for things like this.