Code & Counsel
"Code & Counsel" is a dynamic video podcast where technology meets law. Each episode delves into how digital innovation, particularly AI and machine learning, are revolutionizing the legal landscape. Join us as we explore practical applications, discuss ethical considerations, and unravel the future of law through the lens of cutting-edge technology. Perfect for legal professionals, tech enthusiasts, and anyone interested in the intersection of code and counsel. Brought to you by Quoqo (www.quoqo.com).
For the video version of this podcast, visit https://www.youtube.com/@quoqo5750
Code & Counsel
The Rise of AI in Legal decision - making ( Part 2 )
Chect it out our Youtube channel - https://youtu.be/ln3ZwHl0tEw
Can predictive analytics reshape the legal world as we know it? Join us as we explore how this cutting-edge technology is revolutionizing legal decision-making. By managing extensive data from contracts and financial records, predictive analytics can foresee outcomes ranging from case verdicts to contract renewals. This episode promises to unveil how data-driven insights are empowering legal professionals and businesses to make decisions that go beyond intuition and guesswork, ensuring more strategic and effective outcomes.
Addressing the elephant in the room, we dive into the challenges and myths surrounding AI integration in legal practices. From fears of data leakage to over-reliance on technology, we tackle the common concerns head-on. Learn how custom data sets, understanding AI's internal mechanisms, and continuous model training can mitigate these issues. We also emphasize the irreplaceable role of human critical thinking in evaluating AI-generated information, ensuring that technology enhances rather than diminishes human expertise.
Finally, discover how AI and automation are poised to streamline the justice system, with a special focus on India. We discuss the significant reduction in judge caseloads and the preservation of institutional memory through technology. By eliminating mundane tasks, legal professionals can focus on more critical work, enhancing efficiency and effectiveness globally. Subscribe to Code and Council on Apple, Spotify, and other major platforms to stay updated on the latest in legal tech innovation.
For more insights and discussions on the intersection of technology, law, and business, subscribe to our podcast and stay updated. Connect with us on social media for live updates, behind-the-scenes content, and more. Thank you for listening, and don’t forget to share your thoughts and questions in the comments or reach out to us directly!
Mail us at hello@quoqo.com or visit our website at www.quoqo.com.
Hello, welcome back viewers. I welcome you to the second part of our podcast series, which is called Rise of AI and Legal Decision Making. We'll be having another set of questions to Chetan, who can help us understand a lot of things that we have and we can do using AI. Okay, so could you discuss the role of predictive analytics in legal decision-making and how it might change the way lawyers approach the AI?
Speaker 2:I think this is one of the most fantastic uses of the AI, and you barely scratched the surface on this. And, by the way, there are many movies where similar stuff is used. I don't remember the name of the movie, but there's Tom Cruise right, and there's AI credits that he will murder somebody, and was it Inception or something like that? I don't know. It's one of those things. There was another movie where the AI already knows that you're feeling and it's only trying to shield you from the inevitable end. These are very advanced versions of predictive analytics in play. If you really look at it, if you have a large body of contracts, let's say you're an enterprise, right?
Speaker 2:You may have 50,000 contracts. Does anyone remember what happened or what's going on in each of those 50,000 contracts? It's not possible. The data is also spread out, say, between procurement operations, sales, legal right. And to be able to get this, I mean you have information today which will basically tell you because of metadata, saying when are contracts going to expire, when is the renewal date coming up, and things like that.
Speaker 2:But can it predict which are your failure points in contracts? Can it predict that with a particular customer, this would be the length of contracts that is typically gonna be there? Or if you even load up, say, financial data and you've done this, so based on data, it you even load up, say, financial data, right, and you've done this, so you know. Based on data, it can even predict your EPS for the next two quarters or your predictive sales and things like that. And this is based on real-time data and that can keep changing as well. So, from a very futuristic standpoint, this is like manipulating timelines and starting. So every change in a timeline can impact something else later on and in our case, if you really look at an organization, there may be many things that can have an impact on so many other things as well.
Speaker 2:And sometimes the responses are always reactive and you really wish that you had the prediction ability to be how the outcome is going to be.
Speaker 2:Yeah, and you are a master at this group. I only speak on this from my perspective as legal, but you know, as a data scientist, that some of these things are beyond the realm of science fiction as well sometimes, and there are things that people I know of several startups wanting to do share price speculation and things like that. We'll not get into some of those things, but predictive analytics based on the quality of the data, based on the volume of the data can really transform an enterprise's performance.
Speaker 2:Knowing fully well, let's say you know, ordinary will know that the company has, say, a hundred products and enterprises In an enterprise. You know that two or three of these may be doing well but in reality it may be something else. So from a legal perspective, it's fantastic to have some of these things. You can also predict the outcome of cases, not just because of you know if there's a change in judge or if there's a new element of evidence, then the appropriate weightages given to some of these things tweaked and actually get to a stage where you have you have some reasonable certainty.
Speaker 2:I wouldn't say that it's always going to be. You know there have been too many things and you know to all of these things as well. Yeah, absolutely, but I would. No, I would not, I would, I would, I would defer on that. I think we have to make.
Speaker 2:The more variables you have correct, you know, the more the ability of, the better, the ability of for Gen-AI to juggle all of these things, to control more control points. It's a human's way of thinking, maybe about six or eight data points at any point in time that they can juggle in their minds to arrive at an outcome, but Gen-AI can do 20, 30, 300, 3000, 4000. There's no limit on all of these things. You have to guard yourself against confused data or confused insights. But when it comes to predictive analytics, showcase a trend line for something like this, or let me know when this is likely going to conclude.
Speaker 2:What is the outcome, how other companies have faced something like this, what are the potential fines? Right, things like that become a goldmine, and that today is possible from all of these advances in technology as well. So, and if you have the relevant systems and we do, as a matter of fact, we use this. We have successfully, sort of like, done this as well, and it's just a it's like an insight property of things that you can do. It's very, very useful information. It can also change the way you do business.
Speaker 1:You can decide whether or not to go into a particular you can course correct based on how the outcome might be Exactly.
Speaker 2:So these are things which earlier you would use some very sophisticated minds, your experience or you know there are a lot of strategists.
Speaker 2:You call a board of directors and advisors and a whole lot of advisors to arrive at this right and you know you'll have a debating session and then someone finally takes a call after you know, after hearing everybody saying, okay, this is likely the thing, and you don't know whether it's going to be right or not. It's based on gut feel, intuition and yeah intuition, and there's an interesting book saying how intuition has failed us so many times.
Speaker 2:Think about it, and that is true as well. But here the analytics are based on data points, and the data points are basically what and how you run your business or your law firm or things like this. So there are a lot of uses for these.
Speaker 2:And once again it's like having a key to a lot of exhibit data based on which you can take decisions. If you're a lawyer, let's say, you have more success in family law cases than corporate law. As an example, you can post correct immediately and you can also, as an example, say that you need to take certain calls, say on a particular document, on certain positions. In certain documents, positions are likely to be accepted by the other side. So this is the goldmine of stuff.
Speaker 1:So that's how I view it, but what do you think, Guru?
Speaker 2:Do you think that? Yeah, definitely.
Speaker 1:I think if we can know what we are getting into, what is the risk associated, what are the compliances, different kinds of risks, what is the success rate of a particular action that we're going to take, we definitely will be geared up to adopt an approach that is going to give us more success than just believing in what we are trying to do, or probably leaning on somebody else's experience okay, or their experiences basically. So it's always a better way approach than doing the regular way. I think that on data sciences, or machine learning.
Speaker 2:This would be like the next thing. We've seen the use of predictions, especially in autonomous car driving and things like that, or even autonomous drone flying If it loses Wi-Fi contact, it'll come back to a certain point on the base and things like this. And those are applications associated with different fields. But when it comes to critical fields, like how human reasoning can be benefited or supplemented by the use of all of these things, I think predictive analytics is right up there.
Speaker 1:Another thing is that AI can remove biases. Most of the people who take decisions on a certain course, they are biased or they are not trained to lean towards a certain direction. That's where I can remove those biases, probably clear up your mind and give you ways to think better.
Speaker 2:I think it's true in real life as well, If you take courts and if you look at the cohort of justices in a court, each cohort will have a certain bias. You know, whether it's unconscious or not. It really depends on the person's background, his abilities, his experience and so many things. It's not necessarily a bad thing. Bias is not necessarily bad, but you'll have, you'll get a cohort. Sometimes that's extremely local. You get a cohort. Sometimes that's extremely restrictive. Sometimes you get a balanced cohort. So things like these and it's true even in companies as well I'm saying you know, you'll see newspapers or online and, oh, they're biased, oh, this is happening and it's. You know. One of the ways is to sort of like you.
Speaker 2:The ways is to sort of like you can use AI to sort of like refine this data to make it more even right and take it out such that it's balanced itself out as well. But it's still early days. I wouldn't say that even the data sets that you're using can be biased right, so you have to be able to guard against all of these things as well, and there's no at least today there is I wouldn't say that with a hundred percent accuracy that you'll be able to remove all bias from the system. I don't think that system exists. I don't know. What do you think?
Speaker 1:AI is basically trained by humans, so it is have some amount of bias by default.
Speaker 2:So unless you have a machine, creating a machine which we have no idea, and that's not, at least in today's technology, may not be. And now we are getting into sentient machines and things like that. We're in a totally different Right feel. But yeah, I mean, look, you'll have you know. That's another thing.
Speaker 2:excellent point saying you know you'll be able to at least eliminate bias in some of these things. You know, when you're approaching a decision or you want to eliminate your own bias when it comes to decision making in a particular scenario, these become very, very critical in a particular scenario, these become very, very critical.
Speaker 1:So great Chitra. I think, our viewers are getting a lot of insights on the applications of AI and how it can be used and what are the various use cases, so I also would like to understand what are the challenges associated with AI in legal practices.
Speaker 2:Oh, I have my own personal biases on this because, you know one, I have met people during the course of work where people are like either totally dismissive and the dismissive also happens because they're scared of the changes that are coming and how it might impact them. At the back of their minds, you can basically always gauge that there's fear that's driving all of these things. Then there's also the traditional factors of a few D all around. Fear is one thing which I just discussed. Then you've got uncertainty as well.
Speaker 2:People are uncertain as to how this technology can be utilized and whether they can rely on the technology or not, and despite having's doubt saying, will this achieve mainstream usage? But if you take everything from your AI-powered camera in your phone, on your mobile phone, to everything, you know that it is becoming some sort of it is becoming ultra-basic. It's only a question of time. So, apart from this traditional FUD factors, if you really ask me right, there are a few things that come up saying you know how do I protect against data leakage.
Speaker 2:You can use your own data sets and you can run some of these things in your own systems as well, but keep in mind that if you use a cloud, you can always use or rely on God Rails as well, on God's reigns as well. That prevents abuse or prevents erroneous stuff being introduced into the system as well. If you notice AI, one of the things that the current generation of large legal models also introduces is multi-modality as well as the ability to infer and remember things. So the speed of inference and latency keeps going down and you can basically help and memorize a few things, and the longer you use it, the better it becomes the earlier models you will remember. After five or six conversations, you lose complete track of what you're doing.
Speaker 1:I don't think that exists now anymore.
Speaker 2:And also the ability to sort of deal with deal. At least in the last year or so, the window of tokens that can be processed at any one point in time has moved from 4,000 to minimum to 700,000 now. So I think some of these things are engineering challenges that are being met with better technology, better GPUs, better processing power and things like that. So guardrails and data leakages are something that can effectively be nullified. In today's context, it's no longer a threat, but should be kept in mind If you decide to use an open API to be able to sort of on your data sets. Yes, there could be, depending on the terms of use, to see that your data isn't being leaked as well.
Speaker 2:And depending on whatever LLM that you use, whether you use an untrained, publicly accepted LLM, whether it's open source or not, or whether you have some.
Speaker 2:But, as I would suggest that doing having your own custom data sets, your own mini LLMs, would be a lot better in terms of safeguarding data and also, from what I've seen, there's over reliance. Also on the other end, depending on the generations that are utilizing the technology, newer generations tend to rely much more on AI, while it is generally accurate and generally met to a very high degree. While it is generally accurate and generally to a very high degree, please remember that if you are in an open chat, kind of like a scenario, it basically becomes like garbage in garbage out.
Speaker 2:So you can, you know if you feed it All the biases, all the uncertainty, all the garbage data, you know, you can expect and you get back the same thing.
Speaker 1:Yeah, it's the same thing.
Speaker 2:So you know I would be. So you need to use this in. If you are relying and there's a tendency to over-rely, especially in larger versions of workers please use custom data sets, custom lens. Things like this Deliver to what you require with an SD card. Also, one of the casualties is knowledge about this, traditional human knowledge. It's like saying you know, take an example, how many car makers manually assemble their machines? I would think zero, right? I mean, if you visit the key upland here in India, then the whole thing is run by probably like 20 people, right, as an example, because that is the efficiency in the robotics that's at work.
Speaker 2:Only the robot knows whether it is effective or not? It's no longer a question of human intelligence at work and that's a casualty in knowledge, maybe, as like the law, things like that, but it's one of the things is you rely on the technology to think that it knows, but they should be able to spot some of these things as well.
Speaker 1:So you're saying that we should not use AI as a black box should basically also understand what's going on inside. Yeah.
Speaker 2:I think one of the things there is to also document an LMS performance benchmark. It look for errors and things like that and continuously train it so that over a period of time all these things are illuminated as well. It's only a question of because the reliance on these technologies is going to be a lot more as you move forward. But if you talk to knowledge workers of a previous generation, they will remember all of these things.
Speaker 1:But it's very easy to spot errors.
Speaker 2:So maybe that kind of knowledge may not exist anymore, it may not be required as well, but it becomes the organization's duty to ensure that some of these errors, the benchmarks, the latest technology, the latest models are used for all these things as well. Also one of the things which I, since both of us have kids of the current generation so one of the things is.
Speaker 2:I constantly believe this. You know it is missing. It's critical thinking. You really think about it. What does critical thinking mean? The ability to perceive whether a source of information is correct or not, can be relied on or not. So this kind of critical thinking becomes much more of a human trait, or a requirement for a human trait to be able to utilize AI better, and that's how I would like to think about this. So critical thinking is a much needed skill, and especially if you have knowledge workers utilizing AI a lot, and there are certain professions that do this. Well, google has a lot of hopefully critical thinkers around, but it's one of the skills that is required.
Speaker 2:So I would think that these are challenges of using AI either in a law firm or in an enterprise setting and things like that. But I've also given you the solutions for all of these things as well. Okay.
Speaker 1:So, Chetan, we would like to probably the viewers would like to know how COCO has you know what are the solutions COCO has developed and how it has solved some of these problems. So can you just talk about how focus ai review, about the focus ai feature, how it can enhance the document analysis?
Speaker 2:yeah, I think. See, we build products uh around ai that simplify a lot of things as well. I won't get into every product and what it does, but if you have a document, expect a review in seconds. You don't have to wait for hours or days to get back a response. You get it back in seconds.
Speaker 2:I think that, to me, is a key feature. Especially if you're an experienced person. You want some key points immediately. That's exactly what it provides. Now, if you are a small law firm but you're getting larger mandates, then you use something like Gen-EI. You're adding various tiers of associates without actually hiring those. You can use something like sectoral review, things like these, or M&A to be able to do some of our products, to be able to really add a level of expertise pretty much immediately for all of these things as well.
Speaker 2:If you are an overworked legal department, then expect a lot of efficiency immediately. Think about it. If you have, say, three people, but looking at 3,000 contracts a month as an example, then that's not scalable from a human perspective. But with this it's totally scalable and it's available 24-7. So I think these are some of the things that I would, you know, for me, you know, really work. So just to sum it up once again, it's on efficiency, scalability, ability to add knowledge tiers very easily. It can also be very cost effective. Especially as a business owner, these things really matter as well. So I think these are things that immediately come to mind.
Speaker 1:Okay. So Chetan, let's get a little bit more features you can try to see and understand. How do you see the role of AI in the legal sector heading in the next few years?
Speaker 2:let us say I mean, we've already talked about this initially, guru, but I will just take a step back and look at it say from a justice delivery standpoint. You know, at the end of the day, for instance, in India, there are more than 50 million cases pending. Would you like justice to be delayed all the time? Would you like justice to be delivered on time so that it actually impacts the rule of law?
Speaker 1:things like that.
Speaker 2:So, if you really think about it, we've had collaborative human technology. I wouldn't call it technology, I would call it collaborative. Human endeavors like mediation and arbitration are coming out correct, and these are, you know, as an alternative to case to cases in courts, and they have been partly successful. It hasn't decreased the case load, correct, and it's only become more expensive. Think about it. And justice is not just about the high profile cases or the commercial cases and so on. Think about it. But today you have a system which can perfectly streamline all of these things. You have automation and AI systems which, if used in tandem, can simply reduce that entire caseload to just a couple of years and the whole thing can be sorted out. You only need judges to be able to give the final sign off, because today's technology has the ability to not just emulate, but use all the knowledge and the precedents and everything else right. Whatever you use, decision and stuff, you can do this. You can also take the shape of any legal or juristic person that is an expert in those particular fields as well. What it requires, finally, is to also cut down. What it does essentially is cut down the load on judges.
Speaker 2:If you look at the justice system. Today. It's just now discovered Zoom. Think about it. Because of COVID, virtual court hearings became a thing, and that's how it's been going on, but that technology has existed now for what? 30 years, since the time it's going. Invented it, things like this. It's been there for a long time, but what I'm just saying is today you have the ability to really bring that number down.
Speaker 2:There's a very realistic chance to do this and not at a very great cost as well. And it doesn't mean that you have to hire hundreds and thousands of judges or you have to go through the system up and down and things like that.
Speaker 2:As long as you're able to have experienced people sign off on judgments. After listening or hearing all the points, I think that becomes very critical. Unfortunately, we've had three new criminal justice legislation come out just this week. None of this actually tackles this, because this was probably done before all of these things. But if you're really looking at, say, even the criminal justice system and you're bringing some efficiency into it, especially in India, you will use JNI. There is no option. You cannot put down 50 million cases just because of a whim. It's not something that you can do. You have to use technology to be able to do this. You have to sort of like put it down.
Speaker 1:So you know, otherwise there's no point.
Speaker 2:You know it keeps building. There's pressure, the less respect that the people will have for the justice system because it's not being delivered on time. So these are things that legislation has to really pick up. Unfortunately, if you look at legislation today, they'll say, oh, let's tackle this, let's cut this down because you want to cut it down.
Speaker 2:It really is not the case If you use it properly it can really cut down the excesses of the justice system, especially in places like India, where it's actually not being delivered on time, and I speak to this as a lawyer who has experienced all of these things, and I don't think, someone who has experienced this for two decades. So these are some of the things that I would really things that I would really really address. The other thing which we also talked about, which really matters, is institutional memory. In many organizations, whether it is government or justice delivery or companies or law firms or any other type of entity, right, institutional memory is key.
Speaker 2:Who did what? Who did talk about this when? What happened, why? What is it that's doing? This is what is missing in many organizations. Even you and I have dealt, in the course of ProPro, situations where institutional memory is so absent and you really require it. So there has to be a system or some investments that go into all of these things as well, and that is possible today at a scale which was never possible before.
Speaker 1:Technology that has come in with all the tools that we have today.
Speaker 2:Yeah, absolutely, and it's accessible and it's very powerful. And also, while this technology, the advent of technology third point, which I want to also stress is the focus is not on achieving the smaller things, like filling up farms should not become a big thing. You shouldn't have humans actually sitting and doing all of those things you know. You really think about it. Why would you introduce inefficiency when technologies are available to be able to counter all of these things as well? From a very futuristic standpoint, even today, for that matter, you may have gone across some websites where it has knowledge of what you did, perhaps because you've studied it before, perhaps because you've answered those questions before, perhaps it's led by an AI system. Moreover, it knows what you've done earlier. So these are some I call this form filling human time is wasted on form filling.
Speaker 1:you don't do this daily, regular standard of jobs, basically yeah, and you shouldn't.
Speaker 2:You know this should be. If you really look at it and we go back to the Justice, billy Ed is Justice denied example saying you've got to use these. You know, instead of form filling, you should use this for something else. So this becomes very, very critical and you should use the current generation of technologies to be able to streamline all these things as well. Right? So I think that's where things are and that's how I would sort of like conclude on where do you see or where do you hope that the technologies will go through, especially in the legal tech sector? And you know a lot of our examples have come from our home market, which is India. But you know these are true in all parts. They must be relevant in any country or any region Correct.
Speaker 2:And that is exactly what you should you know in the next five years. You hope that all these things are achieved in spite, but thanks so much, guru, and I really appreciate your time in coming up with these questions and asking. Some of these are very specific to me some of the responses about how I think, about how AI could potentially be used, and things like that, even with our current technology. So we should sort of I look forward to our next set of sessions where we cover all the latest things about AI and impact. Thank you, thanks a lot, Jason.
Speaker 2:Hey, thanks everyone for listening to us or watching this on YouTube. Please don't forget to subscribe to Code and Council. That's the name of our podcast. It's now available on Apple, Spotify and all of the major podcast directories. Don't forget also to press the like button if you're watching this on YouTube. Thank you.
