Most, if not all, legal AI applications at the moment are text-based rather than voice-based. But, now here comes Loop, which offers the general public the ability to use spoken natural language processing (NLP) to find legal information.
The Philadelphia-based start-up got going about one year ago and is this year in the process of coming fully to market. Artificial Lawyer spoke to one of the co-founders, Peter McCann about this ‘Legal Siri’.
McCann begins by noting that at 37 years old he is in fact the youngest of the group that created Loop. He also notes that the NLP algorithms that deal with the speech recognition and processing have been patented by the founders after a period of development that including trials in the medical field.
So, first point, this technology has not just arrived, it’s taken time to get here, even if Loop is new. But why apply this tech to the law? Why not use it in another area?
‘Loop is about the recognition that 80% of people who need access to the law can’t get represented by a lawyer,’ McCann says, citing a now established figure in A2J circles.
‘Most people don’t even know they have a legal issue they can respond to,’ he adds, pointing out that many people simply have their rights impinged, or suffer some form of loss or harm, and because gaining legal advice would be too expensive simply give up and don’t even ask to find out.
‘That’s what we want to solve, this is not the way law was intended to be,’ McCann states.
All good, so far. But, what does Loop actually do? McCann is enthusiastic on this too. Clearly there is a lot more to it than the description here, but put simply Loop listens to the question, perhaps spoken to your smartphone, just as if using a voice-recognising AI assistant, then Loop locates the right ‘channel’ of legal information (family, home, work, business) within its data store. Within that channel Loop presents a list of informative tiles, uniquely tailored to the question and questioner.
From there the user can further ‘loop’ through the system, picking up ever-more tailored information and guidance, but, McCann adds, the company recognises that in many cases a lawyer will have to take over the matter at some point.
The aim he says is not to take away business from lawyers, but rather to help people gain access to justice. In short, while some lawyers may feel a little miffed about a legal Siri on the market, the reality is it could end up feeding the legal market with even more work than it has now.
But, coming at this from another perspective one could then ask: why do this if it then demands a lawyer? McCann responds succinctly to this too: ‘We want to become a gravitational point [for the law] so that that people can approach us first.’
He adds that there will be other legal tech tools out there for people to use and there will lawyers too who are needed. But, Loop will be the first port of call and at the centre of the legal journey.
For now, Loop is going to focus on US family law. Specifically, family law in Philadelphia. This makes sense on several levels. First, several legal AI companies have started out on one main practice area, testing that and developing before moving into new areas. For example, ROSS is known for focusing on US bankruptcy law. And even before the founders moved over to the US from Canada, the initial focus had also been on one practice area, in that case it was employment law.
Another reason is that machine learning needs to study user activity to improve. Starting with one practice area at least cuts down on the scale of information the machine learning has to deal with.
But, to play Devil’s Advocate again, isn’t human speech just too complicated to map over to legal problems? After all, very few lawyers speak about legal issues in the same way, let alone the general public who have in most cases no grasp of legal jargon, or perhaps even have the right words to describe what has happened. This is a problem that A2J bots using written text have found with their users, surely for a spoken system it will be even harder to navigate this challenge?
Now, this is where it gets really interesting. McCann explains that to help Loop come to the right answer, or more correctly ‘to push the users down certain paths’ by matching the user’s needs to a corresponding legal issue, the system tries to tap other information it can gain.
This other information includes a user’s GPS location and other data the person’s device can share with Loop. ‘This helps to refine the experience,’ McCann says, which he adds then makes it easier to ‘funnel’ the user to the right information.
And, the longer you speak to Loop the more it is able to help, because it can identify the legal problem more easily. It can also understand multiple spoken styles.
But, is this a legal Facebook? I.e. a system that invites you in, provides a useful service, but also scrapes plenty of data from you? And is the data only used to help the person with their legal issue?
McCann insists that Loop is in many ways the anti-Facebook. Loop is only interested in PII (personally identifiable information) to the extent the user wishes to provide it for achieving their desired outcome, such as populating forms.
‘Loop only wants data that helps us enhance the experience of getting accurate, tailored resolutions to user questions.’ McCann states. ‘Facebook collects data to understand the individual behind the profile, we do the opposite: collecting data that sheds light on how to provide better information or answers.’
And perhaps even more surprising about this new legal tech innovation is the fact that Loop, which currently has around 20 people working in some way on the product, is not looking to charge a fee, not yet anyway.
McCann says that they are well-funded already and the goal is more about perfecting the system and looking for large investors, rather than looking to gather fees from early users.
To conclude, McCann states that the end of the process is to ask the user: ‘Would you like to take action?’ And if they say yes, then to point them in the right direction. And surely this is a good thing and a way to increase access to justice?
If successful, this could quite realistically overcome some of the obstacles access to justice currently faces and begin a new phase of voice-based legal AI. Watch this space.