Our latest GCRC Interview in our series about startups in the legal market. The interest in legal tech and innovation is huge at the moment as VCs look to seize the moment amid the shifting legal climate.
This interview is with Anna Ronkainen from TrademarkNow.
To find out more about TrademarkNow, you can have a look at their website:
You can also look at Anna’s coverage of legal startups on http://www.legalfuturology.com/
First of all, we know that Artificial Intelligence (AI) and the law go back a long way, in terms of research and academia in particular. Are we now at the stage where we are going to have a real-life breakthrough into artificial intelligence actually being used in the practice of law?
Yes, I think so. In a way this is a problem because when, in general, things start getting useful they start being called AI, so more broadly outside the legal context, you are probably using tools every day in every hour that at some point would have been considered AI. Now that they do something useful like Google search or a spelling checker or something more serious they’re not called AI.
Legal technology has been lagging so far behind other fields that it is reasonable that it will have to catch up at some point or just start making progress to at least reach the stage of the 1990s. But at the same time we can also learn from past mistakes in other fields. I think the problem is much of the research in legal has not really been usage-oriented at all. It’s been very, very, very theoretical in the sense that it doesn’t really lend itself to practical applications. In the past few years there certainly has been considerable movement in the legal technology field to get these more modern technologies into wider use. I’m quite optimistic that that is what will happen of course, but it’s a puzzle.
By practical intelligent technologies, what do you mean by practical as opposed to theoretical?
It’s more broadly a problem that basic academic research study problems for their own sake and not necessarily for the practical use that might be derived. Our work is derived out of some very basic research. It’s about vagueness in the law in very general terms; more specifically that is the process of having to study through through trademarks – that’s how this system sort of came about, or what it’s based on.
So is the reason for optimism for these more practical developments to do with the legal business – law firms and legal departments – in companies, leading the research more?
Certainly the financial crisis gave a big push in that direction. Clients of lawyers, especially the corporations that were earlier quite happy to pay huge legal bills, they are no longer so happy to do so. They want to see efficiency and cost savings which can be achieved through technology.
It can clearly be seen that there are different kind of interests. I guess the rise of the start-up culture also. There are so many legal start-ups – even if many of them do the same thing. I mean there are of course the usual sort of effects of, if one company does something then you might easily get ten different companies trying to work on the same problem. Of course, in the legal field if you have a solution for one particular field, it probably will be more helpful in trying to solve a different field.
For a lot of different start-ups starting at the moment, what do you say is going to allow some of the legal start-ups to succeed and where do you think others will fall by the wayside?
It’s about finding actual problems that need solving and saying what they need. What is the actual problem that needs solving? What does the customer actually want and is willing to pay for? I think that is it really and obviously then having achieved this, the capability to solve the problem, and the will to execute is important. There is no shortage of ideas but then actually being able to do something about them, that’s the big challenge. You often see the figures that 9 out of 10 start-ups fail so definitely there will be many many failures also in the legal field. One has to be optimistic about one’s own though….
Onto Onomatics, as far as I can garner, it is using AI to ascertain trademark similarities using different rules and data. Is that an accurate description and what kind of solution is that providing?
That’s exactly what we are doing. We are approaching the question of trademark searching, or the question of trademark risk analysis, with modern technology because the existing search providers are basically using the same technology that would have been available in the 70s or the 80s. So there are database queries where you just can use wild card searches but you have to figure out those by yourself so the result is that you get search report that takes hours and hours to process to get the same information that our system does in 10-15 seconds.
And on your website it says that the softwares are easy to use. How important is that because obviously lawyers are not always the most adept at using these new technologies?
Based on the feedback we have received its definitely well designed and easy to use. The usability of medical data systems, for example, that’s a huge topic in the usability literature. I have written myself a research article about the usability of legal research systems and that’s the only one I have come across so far. As far as I know there is very little research done on these things. So in the legal context, in order for someone to start using a new system there have to be some exigent circumstances, or it just has to be better because people are accustomed to the idea that in order to start using this new piece of software you have to go take one week course or something like that. Basically when software is difficult to use then obviously it takes up time. We are trying to do our best to make the whole experience of using our system as effortless or painless as possible.
We’ve had a few people say that lawyers are going to have to become more technological as the law itself becomes technological. So do you think that despite these kind of systems like yours and others we are going to get into a situation where lawyers are going to have to become very kind of computer literate?
Certainly there is going to be some getting used to this on the side of lawyers but there is an imperative on the side of the developers of the systems to make such systems for the few of the people that want them in a way that these people actually want to use them. I think that’s the way forward and part of the challenge is that if the system does something very complicated that does not necessarily mean it has to look complicated. A big part of the challenge is to do something complicated and make it look as simple as possible.
That sounds like the opposite of what lawyers are trained to be…
For something like trademark search if you use the traditional database approach, if you want to make something better you add more search options but then you have something that looks like a TV remote control – you have all these options but they don’t really help you towards the policy you’re actually looking for. In our system you just input the trademark and the product and the regions or jurisdictions you are interested in and that’s it; instead of giving you 20 different parameters to choose from without actually answering the question you’re interested in – i.e. whether I can register this trademark – you get an answer.
You’ve just said that the task of technologies is to make things which are complicated look simpler, and as I said, most lawyers are prone to do the opposite. An example of this with lawyers is hourly billing whereby lawyers can get more money by making things more complicated and spending more hours on them. How do you think this concept of hourly billing is an obstacle to be overcome for legal technology and the law as a whole?
The hourly billing model doesn’t really offer many incentives to do things more efficiently but if we’re approaching this from the general or corporate counsel, in-house counsel, then it’s a important to work things out to arrive at a more efficient solution. So that’s an opportunity for companies like us to do something that lawyers are not willing or not be incapable or not properly incentivised to do.
The other thing that we’ve seen developing is combining both legal project management and automation. There’s this whole idea of kind of mapping a whole process genome for example. How close do you think we are to getting to a situation where we’re heavily automated and lawyers end up being more specialist rather than doing the legwork they do at the moment?
This is the kind of question where you are wrong regardless of what you answer really. I think it will obviously happen in piecemeal fashion. We’ll be in a situation where solving one legal problem is not directly translatable to another field – it will definitely not be an even process. There will be some areas like criminal defence that will be the most difficult to actually do anything useful and reliable enough for it to be worthwhile. But in something between 10-20 years is what I think it will take or this to happen…
So not too far off Richard Susskind’s 2033 kind of estimate?
Something like that definitely. I think the progress is accelerating simply because people, especially when they are younger lawyers, are used to using sort of modern or contemporary technology in other parts of their lives so they’ll either exert more pressure onto their superiors to find a more sensible solution, or they’ll act as agents in this change in ways to do things differently themselves.
Because legal education is still very traditional, do you think that more is going to have to change in legal education towards making lawyers more adept at using technologies?
Well it then goes to the whole question of what is legal education is supposed to be about . What we were talking about in my presentation was that there is a need for people to actually able to build these things – these people will need to have an education in both law and computer science and obviously there are a few places like who do this kind of stuff. The places that do anything like this in Europe do traditional AI which is very theoretical and not application oriented.
Onto your academic research, could you help us and our readers to understand what Fuzzy Law is and secondly where is this research taking you?
Traditional crisp logic is about having truth values of exactly one or zero. With Fuzzy Logic you can operate with all the values from zero to one, so you’ll have this whole range of values. Its sort of useful for something like answering the sorites paradox. Removing one grain of sand from a heap of sand does not make it not a heap of sand, it’s still a heap of sand; but if you do it an infinite number of times then you will end up with nothing so at which point does it stop being a heap? The logic is useful in the light of modelling vagueness in the law, so legal concepts that don’t really have a clear delineation between their field of application and the rest of the world. It is sort of it’s a bit difficult to describe…
I’m sure there’s a lot of complexities associated with it but from what you are saying there, dealing with things where there’s no complete one or zero truth. One of the reasons why hourly billing rather than some alternative fee arrangements has persisted is that lawyers would talk about the complexities and contingent factors that will affect a given matter they’re being asked to do. So to that extent, how much can you apply the things you can get from Fuzzy Logic to billing?
You can model anything with Fuzzy Logic if you absolutely want to; it can be a matter of routine to weigh factors on different things, that’s very easy. You first have to figure out what is the problem then start looking for factors for it so you can represent it.Then you have to figure out how we can solve it and have the right tools for it. In our case, it just happened to be a coincidence that I had the model for Fuzzy Logic. With trademark software, strictly speaking we don’t use Fuzzy Logic, but we use something similar; we just use different numerical factors to model different legal features but we don’t in the strict sense we don’t do Fuzzy Logic.