Eliza, how do I claim asylum?
April 7, 2017 8:59 AM   Subscribe

Browder says this new functionality for his robot lawyer is “long overdue”. He told the Guardian: “I’ve been trying to launch this for about six months – I initially wanted to do it in the summer. But I wanted to make sure I got it right because it’s such a complicated issue. I kept showing it to lawyers throughout the process and I’d go back and tweak it.
The creator of a chatbot which overturned more than 160,000 parking fines and helped vulnerable people apply for emergency housing is now turning the bot to helping refugees claim asylum.
posted by MartinWisse (30 comments total) 51 users marked this as a favorite
 
Beep beep boop
All lawyers now compute
Our career's at an end
Thanks to our learned silicon friend
posted by the quidnunc kid at 9:04 AM on April 7, 2017 [2 favorites]


[Judge] Why is your client seeking asylum?
[DoNotPay] I think you ought to know he's feeling very repressed.
posted by delfin at 9:14 AM on April 7, 2017 [12 favorites]


I have done marketing for retail law practices (wills & estates, personal injury law, mostly) and the lawyers I worked with have high hopes for technology and automation. One of the reasons lawyers are so expensive is because the discipline is so labour-intensive. Lawyers or associates have to go through boxes of material by hand. Or lawyers will have to spend time in face-to-face meetings going over paperwork that the client could have done ahead of time.

There are a lot of lawyers who realize that measuring performance by the metric of "number of bill-hours" is, at the end of the day, not providing good service to clients, especially with retail law, where clients don't have any money. Wills and estates is a good example.

And if you're not providing value to your customers does it matter if you, as a lawyer, are going to get less work because of automation?
posted by My Dad at 10:06 AM on April 7, 2017 [7 favorites]


The level of desire to help other people, combined with the massive amount of work that has gone into this is, well, humbling to say the least. Sort of ironic that the first thought I had in considering the significance of it's reach was, "wow... here's a person who deserves the Presidential Medal of Freedom."

Damn.

Mr. Browder, I tip my hat to you.
posted by BigHeartedGuy at 10:08 AM on April 7, 2017 [28 favorites]


The article doesn't address this, but what about issues of liability? It's hard to tell what Browder is saying and what the article is describing, but is Browder holding this out as providing legal aid or just as an automated form-filling service? It says he's working with lawyers in various countries; are they accepting responsibility for these people's cases? What happens if this bot messes up someone's application? Or:
Browder says it was crucial the questions were in plain English. “The language in these forms can be quite complicated,” he said.
What happens if his "translation" into "plain English" misleads applicants? It's good that he's helping people but this isn't something you can "move fast and break things" with and fix as you go along, this is dealing with people's lives. Lawyers have ethical obligations in case things go wrong, but what about this guy?
posted by Sangermaine at 10:20 AM on April 7, 2017 [3 favorites]


The article doesn't address this, but what about issues of liability?

Now we need a legal advice app that helps us sue another legal advice app.
posted by clawsoon at 10:28 AM on April 7, 2017 [9 favorites]


The article doesn't address this, but what about issues of liability?

Just needs a disclaimer: "This is not advice." IANAL, so... don't take that as advice if you want to build your own legal advice app...
posted by My Dad at 10:29 AM on April 7, 2017 [4 favorites]


Now we need a legal advice app that helps us sue another legal advice app.

This would probably be less funny if you and your family were desperately seeking asylum but were denied because a chatbot screwed up your application. Don't worry, though, because the developer is assuring you that will be fixed in the next release.

I find legal matters to be very hard to discuss online because people both don't really understand what lawyers do and like to joke about pretty serious matters more than with other topics. And, while I'm a lawyer, I don't do this kind of law so this isn't some protectionist rant. Protecting client interests, especially vulnerable ones like asylum seekers, is a pretty serious matter. Doubly so in this dark age of Trump where the already-complex field of immigration law has gone mad and been turned into a weapon.

There already exist apps and services like this one to help with forms and that's a good thing, but they also create complicated questions that need to be addressed.
posted by Sangermaine at 10:40 AM on April 7, 2017 [9 favorites]


> What happens if his "translation" into "plain English" misleads applicants? It's good that he's helping people but this isn't something you can "move fast and break things" with and fix as you go along, this is dealing with people's lives.

It doesn't seem to be a problem for tax preparation software, to the point that someone's much more likely to make a mistake following the original forms than they are answering the plain English questions in TurboTax or whatever. What's so great about immigration forms that they can't be simplified while still meeting the requirements of the original forms?
posted by ddbeck at 10:45 AM on April 7, 2017 [6 favorites]


Liability issues are great and all but the reality is if you can't access a lawyer, lack of liability is the least of your worries.
posted by Big Al 8000 at 10:46 AM on April 7, 2017 [22 favorites]


What's so great about immigration forms that they can't be simplified while still meeting the requirements of the original forms?

Just spitballing here, but maybe the fact that if you screw up filing your taxes, you owe some back taxes and interest on them, while if you screw up filing an asylum request, you end up, best case, in a Kafkaesque immigration detention system with fewer legal rights than an actual convicted prisoner would have, and worst case you end up deported somewhere where you're tortured or murdered?
posted by strangely stunted trees at 10:51 AM on April 7, 2017 [8 favorites]


I think this fabulous. If there liability issues, then perhaps a lawyer could do a quick review. Input from those lawyers could be used to improve the programs, and maybe now lower income folks could have get access/get screwed less.

We've had good advice in general, but wound up losing money on what a lawyer said would be a valid claim on issueswith a contractor for a couple thousand, but the legal fees would have made it a wash. My husband was afraid to consult a specialist, because he would be starting out at 10k, and could have potentially paid as much as 25k, but he lost that in wages over the years, and got screwed on his retirement because we went pro see. When he finally went to see a lawyer on retiring, the fella we saw said we did a great job, but there was one little loophole that depended on "good faith".... Lost a lot, because we didn't have it up front.

I think most lawyers are OK, but there are crap lawyers that are greedy, lazy, or incompetent, and where's the recourse there?
posted by BlueHorse at 10:56 AM on April 7, 2017


if you screw up filing an asylum request, you end up, best case, in a Kafkaesque immigration detention system

Yeah, but giving refugee and asylum claimants access to this technology (in theory, at least) reduces the workload on the available pool of qualified legal professionals who are helping claimants right now. In other words, give claimants the tools to start the process, and pair the "chatbot" with actual human experts.

Seems like a good way to help more people.
posted by My Dad at 10:59 AM on April 7, 2017 [4 favorites]


If you don't have access to a lawyer, I have a hard time believing that answering a rewritten sequence of questions would be any worse than the original forms. Consider question 3.A on form I-589:

> Have you or your family members ever belonged to or been associated with any organizations or groups in your home country, such as, but not limited to, a political party, student group, labor union, religious organization, military or paramilitary group, civil patrol, guerrilla organization, ethnic group, human rights group, or the press or media?

This is a 54 word question, excluding the "Yes" follow up, and it could rewritten as 10 or 11 separate, shorter questions and then recombined to fill in the original form. It's not rocket science; it's asking questions in a way that's not practically designed so people will forget to include something important or make serious mistakes.
posted by ddbeck at 11:04 AM on April 7, 2017 [17 favorites]


Just needs a disclaimer: "This is not advice." IANAL, so... don't take that as advice if you want to build your own legal advice app...

At least in my jurisdiction, this sort of disclaimer wouldn't save the founder from sanctions for the unauthorized practice of law or potentially malpractice. Here, the courts have been pretty clear that advising someone how to fill out a legal form is giving legal advice. It doesn't matter if you begin with "hey this isn't legal advice" immediately before giving legal advice.

On the other hand, our court, along with many others, has been very interested in "expanding access to justice," and removing the various barriers to people getting their cases heard on their merits. Seems like this is a step in that direction, though on the whole, the whole regulatory system isn't geared up for handling computers practicing law.
posted by craven_morhead at 11:56 AM on April 7, 2017 [4 favorites]


Browder says it was crucial the questions were in plain English. “The language in these forms can be quite complicated,” he said.

These details are used to auto-fill an application form for either the US, Canada or the UK. “Once the form is sent off, the details are deleted from my end,” said Browder.

The 20-year-old chose Facebook Messenger as a home for the latest incarnation of his robot lawyer because of accessibility. “It works with almost every device, making it accessible to over a billion people,” he said.

Browder acknowledges Messenger doesn’t come without its pitfalls. Unlike some other chat apps, it’s not automatically end-to-end encrypted. Browder says there is, however, end-to-end encryption between his server and Facebook. He added: “Ideally I would love to expand to WhatsApp when their platform opens up, particularly because it’s popular internationally.”
SO much good here, and he's only 20 (!!) - serious kudos and applause for this guy and all who have worked with him.

One thing that irks me: DoNotPay.co.uk doesn't include any info about the service on its website without signing up, but there's another tool that helps you document that you have informed an HIV positive status to a sexual partner
Depending on your location, not disclosing an HIV positive status to a sexual partner is punishable by up to life imprisonment.

Unfortunately, proving that you disclosed is difficult. DoNotPay Disclose allows you to do so for free in under 30 seconds.
These are wonderful services in a world that can be fucking awful.
posted by filthy light thief at 12:38 PM on April 7, 2017 [8 favorites]


I'm with sangermaine, mostly.

What people don't seem to understand is that even now we are constantly trying to fend off "document preparers" who charge naive individuals huge sums of money (relatively) to fill out important forms incompetently. You know what's worse than filling out a student loan repayment plan application by yourself badly? Having someone else do it and charge you several hundred dollars up front plus $39/mo. for the life of the loan to do it badly. For example.
posted by praemunire at 12:40 PM on April 7, 2017 [8 favorites]


I had the opportunity to hear Browder participate in a panel on legal chatbots at the CodeX FutureLaw conference yesterday. He struck me as very thoughtful. He's working closely with subject matter experts (not just lawyers but also charities and aid organizations) and clearly cares deeply about issues such as privacy and confidentiality. He also struck me as very respectful of the opportunity that catching lightning in a bottle with DoNotPay has afforded him. I got the sense that he wants to do this right, and that he's connecting with the experts and resources necessary to go about it.

For example, the HIV disclosure uses the Bitcoin blockchain to allow someone to provide proof of disclosure, but it does not store anything identifying in the blockchain itself. Some (or all?) of the other services do not store any information provided by users; it's all destroyed as soon as the chat session is done. These approaches have limitations (e.g. users can't build a profile or portfolio of documents with DoNotPay), but it helps get around the lack of legal protection that would be afforded a traditional attorney-client relationship.

The DoNotPay services also include options and suggestions to escalate to contacting an attorney. I think that's a very important aspect of any legal chatbot, guided form service, or the like. The service has to understand its limitations and not be afraid to say "your situation is beyond the scope of this service; you really should talk to an attorney, here's how."

The FutureLaw panel included some important and substantive critiques from Joshua Lenon, so Browder and other legal chatbot authors (e.g. the Visabot team) are definitely being exposed to some much-needed perspective on the inherent limitations of chatbots and how they can be done badly or even nefariously.
posted by jedicus at 1:45 PM on April 7, 2017 [9 favorites]


chose Facebook Messenger as a home for the latest incarnation of his robot lawyer

cares deeply about issues such as privacy and confidentiality

Did somebody forget to tell him that Facebook is evil? I'm all for anything that makes legal issues less inscrutable, but having vulnerable people give FB their full legal names, alien registration numbers, and other personal information gives me the willies.
posted by Feyala at 2:01 PM on April 7, 2017 [2 favorites]


At least in my jurisdiction, this sort of disclaimer wouldn't save the founder from sanctions

I was making a joke. It was a lame one, and not that funny, but it was meant to be taken ironically. Ah, this thread brings back memories of working with lawyers, and is also a really good reminder of why it is so difficult for the legal industry as a whole to adopt new technologies that increase efficiency and accuracy and reduce costs for their clients.
posted by My Dad at 2:37 PM on April 7, 2017


At least in my jurisdiction, this sort of disclaimer wouldn't save the founder from sanctions for the unauthorized practice of law or potentially malpractice. Here, the courts have been pretty clear that advising someone how to fill out a legal form is giving legal advice. It doesn't matter if you begin with "hey this isn't legal advice" immediately before giving legal advice.

Are there similar liability problems with publishing reference works about the law? Seems like you could make the case that this is more like a reference work with a very good index than it is like a person giving legal advice.
posted by straight at 2:55 PM on April 7, 2017 [1 favorite]


Seems like you could make the case that this is more like a reference work with a very good index than it is like a person giving legal advice.

Specifically advising someone on how to fill out an application for a legal status? No, you really couldn't.
posted by praemunire at 3:25 PM on April 7, 2017 [1 favorite]


Let me try to explain it this way: the laws restricting the practice of law to lawyers exist as trade protection measures, yes, but also to protect unsuspecting people from the shady and the unqualified in all the important areas of law in which they may interact with the legal system. If one's argument that the vulnerable will benefit from the relaxation of the rules requires the assumption that all or even most of the people who would step into that gap at least be doing so in good faith, one's argument fails. And I think it does, at least in the broad context of apps. It's nice that this guy is at least trying to be thoughtful about what he's doing. From what we already see, I can tell you that the guy selling the $30 Azzzylum!!! app will not be.
posted by praemunire at 3:30 PM on April 7, 2017 [1 favorite]


Specifically advising someone on how to fill out an application for a legal status? No, you really couldn't.

I think you can. The content of the chatbot's programming is fixed, just like a book. No new advice, content, or reasoning is being created in response to the user's input. It's just a glorified flow chart and form template. Nolo and other DIY-law books and software have had such features for decades. Nothing being done here is qualitatively different.

You can point to the magnitude of the possible repercussions of a mistake, but lawyers make mistakes, too. All of this has to be weighed against the fact that this service is widely accessible, whereas for many people attorney assistance simply isn't.

This is not to say there shouldn't be regulation or oversight of these services, just as there is with the legal profession. But I don't think there's a principled objection to the entire concept of a chatbot or guided form filling service.
posted by jedicus at 5:09 PM on April 7, 2017 [4 favorites]


I think you can. The content of the chatbot's programming is fixed, just like a book. No new advice, content, or reasoning is being created in response to the user's input. I

I do not understand how a lawyer who, I assume, passed some form of the multi-state ethics exam which covers such topics can act as if this is an argument from first principles rather than precedent. Roteness is not the determining factor here.
posted by praemunire at 10:03 AM on April 8, 2017 [1 favorite]


There's plenty of precedent: tax filing software, will and trust software, business formation software, and any number of legal self-help books that have been around for decades, many of which address immigration and criminal law issues. All of these have survived UPL arguments.

So, yes, as an attorney and a developer, I don't see how anything DoNotPay is doing is meaningfully different from its predecessors. You may disagree, of course.
posted by jedicus at 10:35 AM on April 8, 2017


I'd also note that the criticism at the conference was "you need to be careful to ensure a high quality service that doesn't mislead as to its limitations." The criticism was not "this is fundamentally illegal."

Similarly, there are lawyers advising some of these services, who have evidently also concluded that they don't amount to UPL. So I don't think the UPL argument is as straightforward as you make it out to be.
posted by jedicus at 10:40 AM on April 8, 2017 [1 favorite]


Oh man, this is awesome. One of the big barriers to running high-value scams over facebook is that people know that nobody in their right mind would put their bank info, ssn, national id number and all their other identifying info into a facebook messenger conversation. Kudos to this guy for doing all the work to break that protective instinct down!
posted by the agents of KAOS at 7:19 PM on April 8, 2017 [3 favorites]


Jedicus, the Ninth Circuit respectfully disagrees with you, at least in some circumstances.
posted by craven_morhead at 9:15 AM on April 11, 2017


Just ducking in to thank Jedicus for mentioning my critique as part of the Stanford Codex panel on chatbots.

Browder and the other panelists where kind enough to take my portion of the panel as the cautionary tale in which it is intended.

Craven_morehead, that is a bankruptcy ruling, a federal jurisdiction with specific procedural admission rules. Most of the other chatbot examples do not operate in the same limited space.

(That being said, UPL concerns can still exists on a jurisdiction-level basis.)
posted by Apollo's Favorite Mistake at 11:15 PM on May 2, 2017


« Older "Obviously, it helps to explain the process to...   |   First cake of the season Newer »


This thread has been archived and is closed to new comments