From 9-11 December, Agami hosted the 2022 iteration of the Agami Summit at the Initiatives of Change, Panchgani, Pune. All three days of the Agami Summit were packed with activities, with performances by folk artists in the mornings and evenings, mindfulness sessions, justicemaker showcases and parallel sessions on different themes during the day. Day 1 also included a showcase of ‘justice-tech’, by participants shortlisted for the Agami Prize.
Glimpses from sessions attended
Considering my interest and area of work in AI and justice, I focused on attending the OpenNyAI sessions. The sessions drew in participants with an interest in technology and AI to solve problems of ‘justice’. Justice here is used in its widest sense, often outside the context of a court.
Key takeaways from the sessions
- Start with automating smaller tasks, instead of fixating on bigger tasks and problems. Thus, adopt an incremental approach to AI.
- Open data systems are key building blocks for AI adoption
- Problem framing is important and to also make sure that an AI solution is the appropriate response to that problem
Presentations by AI experts discussed the opportunities for AI in law, such as in research and education, law enforcement and generative AI like legal briefs and documents, with an under-cutting theme on concerns around the ethics of AI. On the question of bias of AI systems, some argued that, unlike human bias, AI bias can be measured and fixed. Others opined that the scale of AI bias is much larger than human bias. Points were also made on the importance of understanding bias in a scientific sense and the possibility of removal of unconscious human biases through training.
Another point of discussion was the adoption of AI over a period of time in the legal system, instead of leapfrogging with AI solutions. This incremental approach was also considered appropriate considering how outdated the legal community’s understanding of technology is (from a technologist’s perspective).
The OpenNyAI sessions also reflected the ‘open data’ approach for making data more accessible and building registries, which would be the building blocks for AI systems. We saw the launch of the OpenNyAI labs, accompanied with an interesting preview of a ChatGPT modelled interface for queries in Indian languages, made possible with the Bhashini language translation platform.
The OpenNyAI mini-labs offered a nice introduction to the community mentoring and support to innovators who are using AI for solving problems – from making information widely available to detecting abusive language on personal chats. Participants were provided with a roadmap on how to assess problems and the suitability of AI to solve these problems. The cruces of the roadmap to test AI suitability for problem-solving were whether the problem is at scale, a frequent problem, repetitive and/or whether the solution would be at scale. If the responses to these questions were in the affirmative, it could be a way of figuring out if the problem requires an AI solution.
- The Summit offered a window into the millions of things that are happening in the justice space, generally and from a technological perspective. For example, I learnt about CrimeCheck, a service for running background checks for banks using court judgement data involving persons applying for loans. Convenings like this are a good source of knowing what is happening across the country and a reminder that there are many things you do not know.
- We do need to engage a little more critically with the concept of ‘open data’ and what exactly is the data being referred to. Since openness is not a universally value-neutral term, it is a good time to renegotiate terms of the use of data collected from public activities like justice delivery. For instance, as a litigant in an inheritance dispute, would I be comfortable with having details of the property and my relatives being openly available or is openness only associated with aggregate data and not personal information? And who gets to use my data – is it only for research or is it also open for a third party to monetise for profiling and background checks? While registries and interoperable databases are great ideas, they can also make abuse more efficient. For example, by creating a registry of advocates linked to judgements, does it also help build a database for doxing advocates who represent, say, accused of terrorism charges?
- This reflection is perhaps biased because I focused on the OpenNyAI sessions and a community that is working towards building AI solutions for justice sector problems. We should take a step back and assess how we are approaching problems. It felt that the conversation was focused on problem-solving from an AI-first perspective, therefore problems were being framed in a way that lent themselves to AI solutions. While this may not be inherently problematic, it does foreclose conversations on other ways to resolve problems in the sector.
- I also felt that there was a selection bias because of the nature of the sessions, such that differing viewpoints barely made it to the same table, or when they did, there was no scope for facilitating that conversation. For instance, most participants at the OpenNyAI sessions seemed like people who believed in the value of AI and tech solutions, with some reservations, but I do wonder what a conversation on inclusion, bias and discrimination would have looked like with representatives from feminist organisations.
Having been launched a few days before the Summit, ChatGPT made an appearance on Day 2. It was quite entertaining as the tool responded to an instruction on a Shakespearian ode to Brazil’s loss to Argentina in the World Cup. As mentioned before, we saw a preview of a ChatGPT-modelled tool for querying questions in Hindi, Marathi and Kannada. Using a Whatsapp interface, it was tested with participants asking audio questions about government schemes and programmes in these languages, and the tool translated and responded in audio and text in real-time.
The Idea Prize 2022 and Shamnad Basheer Prize 2022 winners and special mentions were announced on Day 2 of the Summit. In the evening, journalist Pankaj Mishra led us through the life of Shamnad Basheer, with his parents and sibling in attendance. A beautifully illustrated notebook, by Somesh Kumar, captured Pankaj’s journey throughout India, meeting the people Shamnad’s life intersected with. It was a nice way to remember a person who now seems larger than life.
Other sessions at the Summit included a reimagining of the justice system through ideas of restorative justice, rehabilitation of prisoners, grievance resolution, law school legal clinics and innovations for persons with disabilities. Sessions curated for entrepreneurs focused on solving challenges within enterprises and impact investing in justice. The art of story-telling was facilitated by sessions on zine-making and hyperlocal networks for solving justice issues.
The team at Agami (supported by an extensive network of energetic volunteers) should be commended for the sheer scale at which they organised the Summit and their ability to bring together such a diverse set of individuals and organisations. I do hope that meaningful collaborations emerge from the relationships established at the Summit.