top of page
  • twitterX_edited
  • LinkedIn

Uwais Iqbal


Uwais Iqbal works as an AI practitioner in the legal tech space for several years. He has experience designing and delivering AI, ML and NLP solutions at leading legal tech startups and a corporate innovation lab. Uwais is passionate about Legal AI, particularly how ML and NLP can be used to reduce manual grunt work for legal professionals, create efficiencies and allow legal professionals to focus on the work they enjoy. He recently launched simplexico, which is a Legal AI consultancy dedicated to serving the AI, ML and NLP needs of the legal sector.


Tell me your journey into the world of AI.


My journey into the world of AI is a bit of a meandering one. I started out studying theoretical physics at university and I picked up coding and app development as a hobby. I really enjoyed coding so I decided to shift to technology.


I was fortunate to start off at Eigen Technologies when they were still a young and small team. As the extraction platform was moving from rules-based heuristics to machine learning algorithms, I took an interest in machine learning. As with any startup role, I got thrown into the deep end and had to learn a lot on the job. It turns out that a lot of the mathematical machinery used in physics is also used in machine learning so my background helped with the transition.


I then moved to Thomson Reuters where I worked as part of the Innovation Lab. We did lots of cool projects exploring user-centric AI and applying AI to Thomson Reuters content. We even developed our own custom architectures for internal data and enterprise use cases. It was a lot of fun but I decided to move to ThoughtRiver, going back into the startup world where we were using NLP to accelerate contract review.


It wasn’t entirely planned, but my meandering path took me to leading legal tech startups and a corporate innovation lab. The common thread had been designing and developing AI and NLP solutions for legal. At some point, I realised that I had been developing the exact same algorithms and models across three different companies all focusing on different use cases. This got me thinking about whether there could be a better way of how AI and NLP in particular, could be provisioned to support the legal sector.


Alongside fortuitous circumstances and this light bulb moment, I founded simplexico - the legal AI consultancy. Just as there are legal service providers responsible for the provision of legal services in the industry, we are looking to be the legal AI service provider responsible for the provision of legal AI in the industry.


I would say my journey into the world of AI has definitely been more practical and pragmatic than theoretical and academic. As a result, I’m more interested in the practical and pragmatic application of AI to help solve problems in the legal domain. It’s been a journey shaped by a combination of deep technical expertise and extensive practical experience. But it’s only just the start!


How would you define AI and why?


I’d define AI as the ability to mimic and eventually scale human expertise for a well-defined and precise task. Most people speak about AI in terms of intelligence but I think this is very loose talk and can create an air of vagueness by granting AI this mythical and magical status. AI can do some cool stuff and give the impression of intelligence but under the hood, it’s all just numbers and matrices.


Strictly speaking, there are two kinds of AI. The first is called Strong AI which refers to an AI system that has reached a level of human intelligence; it can learn, perceive and function completely as a human would do. The second kind is called Weak AI which refers to an AI system that can perform a very specific task like extracting a field from a contract or classifying a clause.


All of the AI-powered systems we have around us today are Weak AI systems - they can only perform a single narrow task. This is just my opinion but I don’t think we’ll ever reach Strong AI. Human intelligence isn’t just something which can be replicated through a machine with numbers and matrices - you need something more.


Just as the Industrial revolution saved us from mindless physical labour, I think the AI revolution will save us from soulless mental labour. The popular narrative around AI tends to speak about how humans will be replaced and displaced by machines and robots. I don’t think that’s true. AI will definitely disrupt more traditional knowledge-based professions like legal, but it will help make legal professionals better at what they do and save them from grunt work; soulless mental labour. AI will do wonders to improve the quality of work for legal professionals.


I honestly believe that AI will help us become more human. AI systems can mimic and scale human expertise so we can go from soulless labour to soulful work. AI will allow us to focus our attention on tasks and problems that require uniquely human abilities and capacities - creative thinking, exploration and imagination. This stuff is needed so we can tackle the big problems facing us like climate change and inequality.


How would you define/explain the status of AI's current capabilities and where those capabilities will go over the next five years?


AI gets spoken about in very vague and loose terms. I like to speak about AI in terms of the actions it can perform. I like to call this the What of Legal AI (https://www.simplexico.ai/learn/the-what-of-legal-ai). In effect, there are a number of actions AI can perform.

  1. Extract - AI can be used to extract information from a passage of text

  2. Compare - AI can be used to compare two passages of text to spot differences

  3. Organise - AI can be used to organise a large collection of texts

  4. Label - AI can be used to label a passage of text

  5. Find - AI can be used to find relevant texts among a large collection of texts using a query

  6. Draft - AI can be used to draft and generate text

  7. Summarise - AI can be used to create a summary of a tet

  8. Forecast - AI can be used to predict and forecast a numeric quantity.

Rather than speaking in unclear terms and using AI in a broad vague sense, it’s more useful to speak in terms of the desired and practical actions AI can perform.


In terms of the next five years, I think AI in a research context will continue to push the boundaries of what is possible. We’ll see ever more massive models that can do more and more impressive feats that will give the impression of intelligence. However, for AI in an industry context, I think we will see more applied solutions for how AI can be used in the legal sector. I also hope to see and contribute to AI solutions developed uniquely for problems in the legal domain.


Has legal accepted use of AI? Why or why not?


I don’t think legal has accepted the use of AI. I think this has broadly to do with AI being shoe-horned into products to attract funding and the extravagant claims made in marketing and sales materials. It’s unfortunate but the hype is real - lots of folks have dismissed AI in legal because of the over-inflated expectations and the underwhelming solutions.


It’s common knowledge in the legal space that adoption rates for AI tools in firms are poor. Most legal tech initiatives and projects have a high failure rate. Compounding on that, AI and data science projects (some report ~80%) tend to fail in the early stages before anything actually reaches production and ends up in the hands of a user.


Without a sharp focus on user needs and an iterative approach to the design and development of AI solutions, the odds are heavily stacked against you. I don’t think AI solutions have been designed and developed keenly enough with the subtleties of the legal profession in mind. In effect, there’s a need to develop a lean methodology for the design and delivery of user-centric legal AI solutions to avoid these pitfalls.


It’s not all doom and gloom though. In a text-based knowledge profession like legal, AI and NLP can be used in countless scenarios to help improve the quality of work. There are opportunities across, legal, regulatory, compliance, law and justice where NLP can be employed to reduce the need for human capital, create efficiencies and improve confidence. It’s just a question of philosophy and methodology. Technology should be in the service of humans not in spite of them. Well-designed technology and AI solutions should be enablers, meeting users where they already are and helping them do better what they already do.


To those wanting to learn more about AI, where do you suggest starting?


Definitely not the sales and marketing material of an AI vendor! I’d suggest that it’s a good idea for everyone to have at least a functional understanding of what AI is and how it works. Just like with computers, it’s only a matter of time before AI becomes a mainstay in our personal and professional lives.


I would say that a good place to start is Andrew Ng’s AI for Everyone course (https://www.deeplearning.ai/courses/ai-for-everyone/). The Age of AI by Henry Kissinger, Eric Schmidt and Daniel Huttenlocher is a good book to get a sense of how AI will come to have a wider impact across society at a social and political level.


There really isn’t anything out there specifically tailored for folks in the legal space. We’ve noticed this gap and have been working on a few things to remedy this.


We’ve put together a Legal AI Glossary (https://www.simplexico.ai/glossary) to help simplify the jargon around AI. We’re also developing a Learning Centre (https://www.simplexico.ai/learn) with simple articles on common questions and misconceptions. We’re also in the process of developing a digital course “A Beginner’s Guide to Legal AI”. It’s a work in progress, but you can join the waiting list here (https://www.simplexico.ai/legal-ai-course).


403 views

Comments


bottom of page