Interview and introduction by Vidushi Marda

In India, what role might artificial intelligence (AI) play in solving the judiciary’s problems? What are the courts’ plans for future uses of AI? Who’s making these decisions, and who stands to benefit or be excluded from these ‘solutions’?

I explore these questions in an interview with Akhil Bhardwaj, former privacy counsel at Agami India, and advisor to the Indian Supreme Court’s e-Committee to implement the “eCourts project.” The role of technology generally and AI in particular forms a central consideration in the Committee’s proposed plans, which are encapsulated in a “vision document.”

I am a lawyer, researcher, and Senior Programme Officer at ARTICLE 19 working on the socio-legal implications of machine learning. The vision document reveals an overly optimistic embrace of new technologies as a compelling solution to complex social problems. In many places, the document oversimplifies the disproportionate impact of technical systems on litigants’ ability to access the justice system, particularly keeping in mind the class and caste distinctions that plague Indian society. The current draft lacks a comprehensive mechanism for scrutiny and accountability that goes beyond Monitoring and Evaluation frameworks that do not lend themselves to agility or rigorous oversight.

India’s eCourts project is still at a draft stage, and now is the time to spotlight these complexities — before the ‘ecosystem’ is built. In this interview, Akhil offers important insights into the inner workings of the Indian judiciary as it works towards addressing problems of pendency (the amount of time it takes to dispose of a case) and accessibility. The views expressed here are his own, and he does not speak for the Committee.

Vidushi Marda: Could you tell me a little bit about your role with the e-Courts project?

Akhil Bhardwaj: I was engaged with Agami, an organization based out of Bangalore. We pick up issues within the law and justice space and we do anything and everything that we can in terms of moving policy, helping bring in new technology, creating circumstances where we can do away with blind spots in existing policy.

The e-courts project was originally kicked off in 2004 and we are now in phase three. In Phase I, the project initially helped with putting in place the requisite hardware for digitization of the judiciary. Phase II mostly focused on bringing in systems and services [for] the digitization of the judiciary. Phase III is where the committee was looking for a thought partner in terms of figuring out what the next step should be.

VM: Could you tell me a little bit about some of the biggest challenges that are confronting the court system in India?

AB: At an absolute thirty-thousand feet level, the challenge before the judiciary is to reduce the amount of time it takes to dispose of a case; so, pendency. And the other big one is accessibility.

I would say that the biggest problem that faces the judiciary from the inside and outside is legacy systems and processes. When I say ‘systems,’ I mean both formal systems of governance as well as technological systems of administration. And when I say ‘processes,’ similarly I mean the procedural rules we are bound by as well as the administrative processes that the judiciary imposes on itself.

COVID was in a way a blessing for the people who wanted to impress on decision makers that there is both the need for change as well as the realistic possibility of effecting the change. Because in some ways, the pandemic forced us to have this country-wide dry run, a pilot project of what the judiciary could look like where there was no possibility of running physical courts. The progress that would have taken us a decade was sped up in a year because there was no other way to go about it.

VM: Could you tell me a little bit more about what you mean by ‘legacy systems’?

AB: Legacy systems here refers to existing ways of doing things. There are several courts which function simply on the basis of the date given by the judge or rather the bench, on the date of the previous hearing. So say we were called to the bench today and at the end of the hearing the judge or the bench will simply pronounce that the next date of hearing is going to be the 15th of January. So, if a particular date is being granted by the bench, the bench will often be able to take into consideration, have [we] already assigned a heavy matter to be called on the same day or not? Is there a full bench assembly on the same day? Are there cases on the days preceding that are likely to spill over into this date?

Now, a judge cannot have all of this data in his or her head. They are not an administrative manager. So, even though to some extent they can mitigate these harms, machine learning can do it way better. The fact that we do not use an intelligent scheduling system is alarming.

VM: To what extent is this framed as a problem that can be solved by technology, and to what extent is this a problem that needs to be solved though overhauling existing institutions?

AB: I would like to talk about how we have conceived of the issues that are plaguing the judiciary. One is the creation of a platform infrastructure, the second is process re-engineering, and the third is putting in the right infrastructure support. And we believe that in the absence of the other [two], none of these three is going to be able to address the problems that we are trying to solve in isolation.

The first one is platform architecture. The first phase [of] the project brought in rudimentary hardware, ensured network connectivity, started digitizing some case records, bringing in some case management systems and I believe hardware was installed across 13,500 courts. And here when I say ‘hardware’, I mean computers, printers, Wi-Fi or LAN systems.

Then the second phase created digital services. The fact that we started having daily orders and judgements published online was a product of phase two. Several courts now have an e-filing system. By virtue of their design as these monolithic systems, they are not interoperable. They are not evolutionary because if you want to improve on the e-filing system or the e-payments system, bring in intelligent scheduling, or allow for the automated information extraction from the documents that are filed before courts, all of these cross-cutting design segments do not function.

The second prong of our approach is process re-engineering. The civil procedure code says that you need to have a document that is self-attested, notarized and submitted in triplicate before the court. Today e-signatures are already recognized in Indian law under the Evidence Act. So, when I e-sign a document, when I want to file this document in court, what I am really doing is printing out and e-signing the document. And after that I am going to sign above my digital signature on that document because a notary cannot attest to the fact that they have seen me sign that document digitally. So, they need to see me physically sign over an e-signed document and then they are going to notarize that. That, of course, completely defeats the purpose of an e-sign document. So, when I am talking about process re-engineering, we are talking about updating our processes to be in sync with what technology allows us to do.

The third one is having the right institutional set-up. Who is really leading the technology adoption by the judiciary? Are they adequately staffed for it? Do we need a Chief Technology Officer as a part of the judiciary as well, or do we need a judicial cadre that is particularly focused on technology?

VM: Could you tell me a little bit about other kinds of AI technologies that are envisioned [for use in the courts]?

AB: The need of the hour during the initial stages of the pandemic was live telecast of court proceedings because it’s a constitutional guarantee that courts are open by design and a litigant must be able to witness what’s happening in their cases. Not just litigants, but, say, [the] press. We worked with the judiciary to bring in the live streaming rules which I believe were published in October of 2020. The live-streaming rules also spoke about creating live transcriptions of proceedings. We are waiting for an approval of the draft rules.

Article 227 of the Indian Constitution guarantees each high court its role as the administrative head of its jurisdiction. Each of the chief justices of the High Courts is involved and it is within their discretion whether or not to adopt a model rule that has been put out by the Supreme Court e-Committee. Something like live transcription we can immediately adopt. Other sorts of technologies that we are trying to bring in include automated review of e-filed documents.

The systems that we have been so far developing have been monolithic. The moment you have systems that talk to each other there [is] so much more you can do with the information that’s coming in. And, that’s why we keep talking about the fact that pendency is not a product of having fewer judges or having judges that are overburdened.

VM: I am interested in procurement processes. Who gets to provide [these technologies] to the judiciary? How are these technologies assessed? Have they been tested or has the committee kind of looked into how they function on an ongoing basis?

AB: Our role within the committee was to really play a thought leader. What we did advise them on are things like, how do you need to change your current procurement model? Or what are the problems with us adopting individual procurement models where each high court goes and identifies a vendor for their own systems?

The role we did not adopt is to go in and actually help them vet individual vendors for particular systems. What we are currently suggesting is that neither we nor the judiciary has the competence to be making those decisions. Because, what we’ve said is that to be able to make decisions on a platform architecture design level, in view of Article 227, you should advocate for high courts to identify their individual vendors so long as they meet standards and specifications that are laid out by the Chief Technological Officer. I would say with respect to phase three of the e-courts project we aren’t there yet in terms of identifying vendors to actually execute. But what we are saying is that the person who needs to lay down the rules of the game is not necessarily a judicial person. It has to be a person who is going to be answerable to the judiciary so that judicial oversight is maintained. But it has to be somebody who has technology as their forte.

If we take up the role of helping them identify the vendors, we would risk going back into the trap of adopting a technology that is not interoperable. Because a particular piece of technology or a vendor’s proposed solution might work really well for the problem that we are seeking to address right there but it may not plug in with the larger technological ecosystem that we are trying to create.

VM: Has the committee interacted with judges, clerks, or lawyers to get a sense of how they would adopt or react to this technology?

AB: Across the spectrum, we’ve spoken to judges at different levels of the judiciary. We’ve spoken to court administrative staff as well as judicial staff, advocates, litigants. It depends also on how long the person in question has interfaced with the judicial system. So if you speak to a lawyer who has been practicing for more than 20 years, they will usually talk about how having orders and judgements available on the websites of the courts has completely changed practice for them because they get the order so much quicker than applying for a certified copy and waiting six days to get it.

The administrative side of the judiciary is already benefiting from some of the machine-readable case records that particularly Kerala has started. A lot of the information that is extracted from the e-file documents is populated automatically on the case management system and it minimizes the chance of error as well as reduces the time spent by the administrative side of the judiciary in inputting that data. So people have had a chance to interact somewhat with the technologies that have been coming in so far. Though of course right now they are very rudimentary over, say, Zoom and WebEx and the proprietary National Informatics Centre software. But it is the first experience [people are having] with digitally-enabled hearings. Different people have had different experiences and there has been a lot of push-back as well.

VM: Which historically disadvantaged groups have formed parts of the conversations related to the e-courts project? How do you think the use of technology will affect their experience of the courts?

AB: In the document, in the section on the ‘Monitoring and Evaluation framework,’ we talk about the fact that we want to have a shared infrastructure, we want to create active partnerships and adoption amongst networks. At the same time we are looking at how that impacts adoption by dissimilar groups. So, this is not a one-time process that is going to happen at the envisioning stage. The document is very conscious of the fact that this is something that is going to have to be done continuously over the course of phase three of this project, to look at the scale, look at the speed and look at the sustainability of the adoption and its impact in terms of what it really translates to particularly for groups that are currently either excluded from formal processes or in are any way disadvantaged in terms of their accessibility to it.

In India’s history, we’ve never before had a consultative process of adoption of any judicial program. The fact that the envisioned document for phase three was put out for a public consultation was a first in the history of the Indian judiciary. So, I think it’s a good start that we have made in terms of walking the talk of saying that we want to design for accessibility and inclusion. And the next draft of the document should reflect what we heard from people.

This interview was commissioned and edited by Shazeda Ahmed.