AI and Coding for Language: A chat with Ben Alarie, CEO and Founder of Blue J Legal
In my last post, I talked about the way that legaltech was developing in different markets, and suggested that the difficulty of coding for multiple languages and jurisdictions was partly responsible for the uneven applicability of legaltech across the world, with the English-speaking jurisdictions served better than others. I called this the Language Problem.
A few weeks ago I sat down (on the phone), with Professor Ben Alarie, who founded legal start-up Blue J Legal in 2014, after IBM Watson’s challenge event at the University of Toronto. Blue J Legal has just launched its AI-powered tool Tax Foresight into the U.S. market.
Is there a language problem with AI or am I imagining things?
You’re not imagining things. Accurate data are essential for AI systems, and in law, the key data are words and language. If you want to develop a legal solution that is truly multilingual, you need a lot of data from many different languages.
But I don’t think this is just a legaltech problem or just an AI problem – if you think about it, it’s actually a legal problem full-stop. Language is a problem with law and the development of law more generally. It’s true that humans can learn to speak multiple languages but you still have to overcome the friction of switching from one language to another. In some jurisdictions, like the States, that’s not such a big deal, but in Canada [a bilingual jurisdiction], it is often a challenge.
A good example is at the Supreme Court of Canada. At the Supreme Court of Canada decisions have to be published in both languages. When I was clerking, every judgment took longer to come out than it otherwise would have because an English judgment had to be translated into French and vice versa before it could be released.
And look at the UN. There are six official languages in the UN. Every official communiqué has to be translated into each of the six official languages. From this it’s obvious that law’s reliance on language has a major impact on the efficient practice and operation of law.
So there is a language problem in legaltech, you’re absolutely right, and it’s a manifestation of the way the law works – because words make up the law.
How are you dealing with that problem at Blue J Legal?
As a start-up, you have to begin somewhere. Typically you start with the minimum viable product (MVP) – and it really is that, minimal. So you start with one language. And then when your offerings come to market and you have customers liking them and using the platform and making suggestions, you get to a point when you have to accommodate the different interests of your customers.
For example, as we have started marketing to the [Canadian] Federal Government, we have realized that we will need to have a French version of Employment Foresight and Tax Foresight, and we plan to have those released by end of the year but it takes a substantial amount of time and investment. We’ve done this already with HR Foresight – it now works seamlessly in both languages, and that was a ton of work for our team.
We have an internationalization function built into the platform at a fundamental level, but it still takes a lot of effort to add a language, and now we have to do the heavy-lifting to translate the other two products into French. But it’s also worth it, because it means you’re enabling a whole new community of users to access these legal insights in a way they weren’t able to before.
The fascinating part in overcoming this challenge of language accommodation across English and French is that it enhances communication about the development of law in the English and French jurisdictions. For example, the Income Tax Act has different provisions in French- and English-speaking jurisdictions and English users will be making arguments based on the English provisions and French users based on the French provisions but by putting all of that together you’re enabling conversation across both in a way that wasn’t possible before. It facilitates cross-cultural learning. I think the key thing to come out of this will be an acceleration in the development of law.
In how many different markets do you have customers? What languages do they speak?
We are mainly in one market - Canada. We have customers in every province and in the Federal Government.
We recently launched the US version of Tax Foresight and are actively growing in the US.
Canada has been an incredible starting point – we moved from the MVP stage to product-fit, and we now count amongst our customers nine of the ten largest accounting firms, eight of the largest ten law firms in Canada, the CRA [Canadian Revenue Agency] and other parts of the federal government. We have significant reach into medium-sized firms and even some smaller firms who see the benefit of bringing in an enterprise-grade AI solution.
About a year ago we started building the U.S. tax product and expending a lot of energy going through the process of really mastering that whole body of law, wrestling with it and shaping it into platform to get ready for a new jurisdiction.
How are you approaching your launch into a new market?
Since we have such strong relationships with the big four accounting firms in Canada, it has been pretty seamless to approach the U.S. counterparts of those firms as well, through our contacts in Canada. So the U.S. accounting firms are all aware of us and we are currently in discussions with them. We are also approaching the U.S. Government, but we expect that sales cycle to be somewhat lengthier.
What is your biggest barrier to entry for a new jurisdiction?
Training our algorithms on new bodies of law.
The platform is jurisdiction- and area-of-law- agnostic, similar to some of the other legaltech providers. Ours is different, though, in that we’re really trying to give a sense of what the courts would do on the merits as driven by existing case law, and that requires us to collect the existing case law and analyze it using our algorithms on the platform. This requires an investment in collecting, analyzing and training the algorithms on the case law. The nice thing is that our platform is able to accommodate it; but we still need to do the heavy lifting on the training side.
Do you think that non-English speaking jurisdictions are currently underserved by legal technology vendors / products?
I think everyone is underserved by legal vendors and products, so yes!
What do you think the future looks like for the industry? Do you think consolidation needs to occur in areas where there are a multitude of players in the market, like contract management and e-Discovery?
I think the best way to make predictions about the future of legaltech is to look at other areas of tech that are more mature. You can see that there are some advantages to standardization, so I understand why people would make claims that there should be consolidation.
I think it will happen naturally as the market develops. There will be dominant players who see that the best way to grow is to absorb smaller players with complementary products, customers and talented employees. We’ve certainly seen it in tech more generally – think about Amazon, Google, Microsoft, they acquire smaller companies that are innovating and developing new technology and then they make changes to adopt those technologies into their platform and they evolve that way. That is already happening somewhat quietly in legaltech and it will continue.
The real question is – who will be among the last few players standing?