The Legal Tech Ecosystem | AI in Action 2023

Hello everyone. Uh, my name is Colin Es Levy. Uh, and for those who don't know me, let me give you a brief little introduction before we, uh, begin here.

Uh, I'm the editor of the Handbook of Legal Tech and I also recently published a book called The Legal Tech Ecosystem. And I am here to be your dedicated guy to navigating this inate landscape of technology, of business, of people, um, and of ai, which has suddenly thrust itself upon the world and into our lives. Uh, for those who perhaps know a little bit about me, I am frequently, um, asked to go on podcasts and give interviews to talk about legal tech, my thoughts on the industry, uh, my thoughts on how best to sort of evaluate your needs when it comes to finding tech, uh, and or how to kind of evaluate, uh, potential use cases for artificial intelligence.

Uh, and my goal here today really is to sort of illuminate the path we're on with respect to AI and the path that I think we'll be on going forward over the next six to 12 months and, and beyond. And a bit of forewarning, I'm likely gonna be wrong about this because AI and technology tends to wrap up the change. Uh, and I'm not particularly good at giving predictions, but that's being said.

Hopefully some of what I will be saying will resonate with some of you with respect to artificial intelligence and as it evolves, um, and continues to play a bigger and bigger role, not just in our professional lives, rather personal lives as well. So I think it's important to first understand where we are sort of in, in the grand scheme of things with respect to artificial intelligence. I think that, you know, let's set ourselves, sort of, give ourselves some context here.

So artificial intelligence itself is not really a new thing. Elements of it have been around since the sixties and seventies. I mean, considering in the US we landed people on the moon, um, which I will not, we have not been back to, but that's a whole separate thing.

Um, that could not have happened without our ability to compute and analyze loads and loads of data. And that essentially is what artificial intelligence still does. The difference is that now with the rise of generative artificial intelligence, we're now able to interact with these solutions the way that two humans would likely interact with one another, with a few little caveats in terms of providing more specific instructions and or sort of more details around exactly what we're expecting to, um, have as a, as an output.

Uh, but that is all really just a way of saying that artificial intelligence has been around for some time and only now is it sort of seemingly thrust upon us primarily because it's just so accessible. It's everywhere. Uh, chat, GBT bogged, uh, Microsoft being now has built in chat, um, functionality.

And so really, you know, it, it, it's opened our eyes to what is possible with artificial intelligence. And more than that, it's allowed for us to rethink exactly what it is that we do every day as professionals since now artificial intelligence can help us with so many of these different tasks. Um, for example, artificial intelligence can help us with, um, analyzing a boatload of data, for example, around, um, case law.

You know, I suppose you have a litigation matter that you are thinking about, uh, litigating. Well, now you can use artificial intelligence to not just determine kind of what your best line of argument is, but also evaluate the likelihood of that argument being successful based off of prior case law and data and the context of going up before a specific judge, meaning that there's data now and how that judge has ruled on different issues and how that judge has treated different types of arguments. And we can now use data to help drive that sort of decision making, not just through analyzing data, but also through artificial intelligence driven analytics that can tell us sort of how best to craft our argument, whether it's even worth us pursuing this litigation or not.

And so that's, I think, very exciting. But just one example of where artificial intelligence really shines because artificial intelligence at its core is essentially, uh, data analysis. It's, it's basically using boatloads of data to then provide analysis and draw trends and draw out potentially actionable ways to move forward.

And that I think comes, is really important to understand because artificial intelligence solutions at this point are really only as good as the data that is used to train these models. Meaning that, you know, if you have some really sort of niche area that you are returning to artificial intelligence to help you with, it may not be of much help to you because it may not have the access to that particular data and or being trained on that particular data set. You know, when you think of law, for example, you know, generic artificial intelligence solutions like chat GBT may not necessarily be the best because it doesn't have access to some of this legal data that may be more useful and provide more pertinent, um, analysis and conclusions, uh, than, than something that is trained on that specific lead legal data, which is why we're seeing more and more, uh, legal tech vendors use data to develop their own artificial intelligence solutions that then are offered as a subset of a larger set of functionality, um, that they offer through their platforms.

So I think, you know, really it's important to kind of understand that artificial intelligence here has applications everywhere. Um, but at the same time, I think it's also important to see artificial intelligence as what it is, meaning that it's not really, um, here to replace you. It's really here to augment you.

It's here to augment how you work and why you work and how you work. Uh, and that's really where, where it comes into play, um, across a variety of subjects, whether it's contract management, legal research, analytics, uh, whatever it is, it's here to kind of make that very detailed and very time consuming work faster, easier, and cheaper. Um, and so I think really at the end of the day, artificial intelligence is not something that we should be scared of, but something that we should embrace.

And I say that not only from the standpoint of viewing it as something that is here to augment our own intelligence and our own operational decision making, but also because it does not necessarily think on its own. It is reacting to inputs that you give it and producing outputs. So we really shouldn't be thinking of it at this point in terms of, you know, for those who have seen the Terminator movies as perhaps Skynet, um, it is not there yet.

It is not there to kind of view us as a threat or even view anything as a threat. And we're a long ways away from that. Will we reach that point?

I think I'd be silly to say we will never reach that point, but will we reach that point in time soon? I don't think so. Likewise, there's been some talk of perhaps computers being now being able to talk back and forth with, in other words, you know, at some point in the near future, perhaps there being a computer that you can kind of talk to and will do things for you in response.

We're not there yet either, because the problem is it need, you need so much data to be able to train these models to be able to respond to human level thought, human level analysis and human level kind of thinking and consciousness. So we're definitely not there yet. Will we reach that point at some point?

Yes. But as, as many have shown, including Ethan Molik, who is a professor and has posted quite a bit on large language models in artificial intelligence on LinkedIn and on us plug, we're really not as far along in this AI journey as, as some may lead us to believe. And I think that's important to understand because artificial intelligence, while it is everywhere and seemingly growing stronger by the day, uh, we're really at the very start of this journey.

It's really early days with respect to where we are in the artificial intelligence journey. We're kind of at the tip of an iceberg that is very deep and very wide. And so I think, you know, it will get more exciting.

Things will evolve, but we're really kind of at this early stage of it. And so at that, which brings me to my next point, which is that when using artificial intelligence, the human element cannot be overstated. Humans are still needed to double check the work to ensure that the outputs are, are valid and are usable.

Uh, we've seen various instances of where that has not been the case and that has led people as strict. For example, there have been a number of lawyers, uh, who have notably used generated AI to come up with, um, case law or analysis, uh, but they didn't check the work of the artificial intelligence and they just simply used it submitted to a court and the court called them out on it and said, Hey, these cases aren't real. You know, you didn't acknowledge that you used generative ai.

And so that's less a function really of the technology not working properly. The technology did what it was asked to do, but you as the lawyer weren't doing your job in terms of being competent enough to check the work that was being performed and ensure that it was providing with the work you needed and that was gonna be valid. And so I think that's currently a struggle within the courts with respect to ai, that AI is, um, not capable yet of acting entirely on its own with respect to certain areas, as I mentioned, including legal matters.

And so that means that we all just need to be more aware of the limitations of artificial intelligence and be double checking the work and be confident enough and confident enough in our own abilities that can double check the work that is created by these artificial intelligence solutions. I also think, moving on that, you know, where are we going from here with respect to artificial intelligence? I think there's a couple things to keep in mind here.

One is that artificial intelligence is but one of many technologically technological solutions that exist in the world, and it may not always be the answer to everything. And that comes, that really comes down to understanding what problem it is you are trying to solve. Because I think there's a tendency now with artificial intelligence to somehow think that it can just solve any problem you throw at it.

And it's always gonna be the solution. And that's just not the case. It's not the case with ai, it's not the case with any type of technology.

You know, AI can't fix a people problem. For example, suppose you work for a company and you have a bunch of people who, um, aren't doing their job, aren't very happy because they feel like they're not respected enough or their work isn't valued enough. Artificial intelligence can help with that.

That's a people problem. That's a culture problem, which brings, which sort of reinforces the point I'm trying to make, which is that when you're trying to evaluate potential solutions, including ai, you really wanna know what problem it is you're trying to solve and then use that awareness then evaluate potential solutions. And as AI grows stronger and as it continues to evolve and be able to do more things, including more heavy data analysis calculations, for example, there was an article recently about, uh, some, a AI solution being able to solve a math problem.

And people thought that that somehow meant that it was becoming, you know, ever more aware of itself. It wasn't. It was just doing what it was asked to do and therefore was able to sort of think through what the problem was in the context of it had the data to come up with the output.

And so I think, you know, when it comes to figuring out where we're going, I think we're gonna be going more down the line of being able to solve more problems on its own. But that still means we have to define what problem you want it to solve for you, which goes back to understanding what problem you're trying to solve and whether AI can solve it for you or not. Uh, it can't be your therapist, it can't be, you know, your counselor, your boss, you know, we're, we're nowhere near that point.

What it can do is just make what can otherwise be time consuming and repetitive tasks, uh, less cons, time consuming, less, um, repetitive and save you times to can spend more time on things that you, that bring more value to your role and bring more value to your company as well. In addition, I think we're gonna see a greater emphasis on the need for there to be an ethical framework around how AI is regulated. President Biden came up with an executive order with respect to laying out some basic principles around how to govern AI and how to evaluate.

Uh, this was in the context of federal agencies, but I think it's emblematic of this larger movement to acknowledge that artificial intelligence A isn't going anywhere and b is going to need some guardrails around its usage. Uh, the EU came up with the AI Act, which is another effort to try to provide some guidelines around its usage. And the California um, court system also came out with guidelines around lawyers and their usage of ai.

And I think that tho those ethical efforts are super important and are gonna become even more critical as a AI continues to grow stronger, grow more impactful, and be able to do more for us, um, in ways both that are expected and unexpected. Which I think also means that going forward as AI evolves, we're gonna need to be open to continue to experiment, continue to iterate and continue to explore how AI can help us, how it can't help us, its limitations. And also understanding more about exactly how all these different solutions work, which is another part of this ethical discussion that I think is super important, which is sort of understanding what data is used to train these models, where that data is coming from, whether data you're submitting to these models is being used to train or not open.

AI got into a little bit of hot water a while ago with respect to not being open about what data it was using to train its model and perhaps using data that you were submitting to train its model doesn't do that anymore, but it at one point it was. Um, and that's I think another important element of evaluating any AI solution is, as I said before, AI lives on data. So when you submit data, you want to know, is this data actually gonna be seen by a human?

Is it gonna be used to train that model or is it just gonna be used for that one time of having the solution, the AI solution come up with an output based off an input and that's it, it's not gonna be remembered, it's gonna be forgotten. And that's where data privacy is gonna come increasingly into play. 'cause we already have a number of different privacy frameworks around GDPR being perhaps the most well known around what the California Consumer Privacy Act and some other laws as well and other jurisdictions.

Uh, but those laws are not gonna be, um, I think enough we're gonna need to continue to evolve those laws and perhaps come up with more to help sort of constrain the power of ai, but in a way that also allows for its continued evolution and continued evaluation and continued use. And that's gonna be a huge struggle I think for governments, uh, state, local, federal, wherever, uh, because technology is not going anywhere and it's also not slowing down. If anything, it's speeding up in terms of its level of evolution.

And so rather than, I say this all the time to many people, rather than trying to keep pace with the pace of technological evolution, we should just see where we are in comparison to technology. In other words, always have it in sight in front of us 'cause it's always gonna be in front of us, but, and don't lose sight of it. And if we wanna remain relevant both as professionals, uh, and as people, I think in this tech enabled world, that means we're gonna need to continue to adapt and continue to grow and evolve with the rock, with the ongoing evolution of artificial intelligence.

Uh, and that means opening ourselves up to experimenting. That means being open to being uncomfortable. In fact, being comfortable being uncomfortable.

Because you know, if I'm being honest, you know, many years ago when I first had my first taste of technology, I found it intimidating. I found it. If you're inducing, I was scared of it and now I can't get enough of it.

But growing forward, you know, especially as people who have been existing in a certain reality for a long time and thinking that that was the reality, I was always gonna exist. It can be scary to have to dramatically adjust sort of the wor you know, your level of expectation, how you work, given the rise of all these different technologies that can be scary. But again, it comes down to seeing technology as something that is here for you to augment how you work rather than replace you.

And here make your life actually easier and perhaps allow you to spend more time doing things you enjoy, spending more time with families, friends, whomever, rather than doing this repetitive work that now can be done by ai. And I think that's a really important lesson to keep in mind going forward, is that artificial intelligence here is not here to replace you. It is re redefining in many ways some of the work and some of the tasks that we do, but is not here to replace you and whole or in part is just here to reimagine kind of what it means for you to be doing certain jobs.

And does that mean that certain jobs going forward may not be what they once were? Absolutely. Does that mean that there will be jobs that may not exist going forward?

Probably. Which is why again, it's important to be open to adaptation, open to experimentation and open to seeing where things go going forward. And, and finally, I think with respect to AI and where it will be, you know, in the next six months and next year, I think what we're gonna see more and more of is more integration of AI being able to do writing and image work together more and more.

In other words, being able to integrate imagery as well as text together and being able to do more and more sort of ad hoc calculations, whether it be solving a math problem, whether it be solving some other type of problem with less information needed. And what I mean by that is you're gonna have to, you're not gonna be able to have to provide as much information that you used to to an as solution for be able to figure out exactly what you're asking it to do and for it to do it for you. And what really I'm getting at here is that, as we all know with using a lot of these solutions, we have to prompt the solution, meaning we have to develop a prompt that some kind of instruction for in terms of what we want it to do.

And people have written books about how to prompt and there's even been talk about the creation of prompt engineers. I don't necessarily see that as a long-term job. I, because I think that artificial intelligence going forward is gonna be able to be able to think a little bit more on its own, so to speak, and be able to understand what it is you're asking it to do without you needing to be so precise and prompting.

And does that mean prompting is gonna go away completely? No, it's not gonna go away completely, but what it is gonna mean is that you will have to do less work to interact with these solutions and these solutions we're able to provide you with more accurate results with less work on your part, which is gonna make it even faster and more better at augmenting kind of what you do on a day-to-day basis. I could be wrong about that, uh, but I have a feeling I may not be as to where we will see AI going forward, um, after, you know, after a year or two.

It's really hard to say. I think that clearly we've seen, um, a move towards the commercial commercialization of ai. Um, and that was in part what led to the whole OpenAI soap opera with the board dumping Sam Altman and then bringing back on board.

'cause Sam Altman has been very clear and very forthright about wanting to commercialize his work in o in AI and others on the, on the then current board, were concerned about there being perhaps a little, you know, going too fast. But the cat is outta the bag. The cat was outta the bag when chat GBT came to play.

Uh, that cat was outta the bag when people started paying to use a more advanced version of chat GBT. So I think that the commercialization thing is gonna continue to increase. And that raises, I think a few concerns that I'll, that I will reiterate now, which include the fact that a outputs are only as good as the data that's being provided as inputs B uh, artificial intelligence getting better and better replicating human imagery and human voices.

So that means we're gonna have to be more and more careful about how we go about evaluating voices we hear and images we see, which I think will require there to be, uh, a stronger ethical framework and and regulation around use of AI for what purposes, especially with respect to marketing or business purposes so people don't feel like they're being led astray or being misled, which is all too often an occurrence with the heavy rise of misinformation online. Um, and so I think that's gonna be a major concern going forward as well. Uh, and and finally the last thing I would say with the, so to sort of where we're going with this AI journey is that will computers become more and more sort of aware in terms of what you're asking it to do and in terms of how it interacts with you?

Yes, I certainly think that will be the case. I, you know, and we've seen that already, you know, Google search now when you type something in, it figures out what you're asking it to search for. And that's some thanks in parts to AI and computers, likewise, like your phone or your desktop computer or laptop will likewise be better at performing ad hoc tasks.

Thanks to the thanks to AI and ai, AI will be playing a bigger part in the backend of computers. I think we're also gonna see a closer integration of AI into others, into more solutions as opposed to a standalone, um, solution. So, you know, a lot to unpack there.

And I realize that, you know, I've been talking for a little under 30 minutes of this point. There's a lot, a lot going on. So the best advice I can really give to any of you is to stay on your toes, continue to read, continue to follow people, like eat the Molik and myself to learn more about the space and experiment, use all these different tools, try them out, see what works, see what doesn't work, uh, you know, learn and iterate.

And also, and last but not least, you know, there's been a lot of research and a lot of papers that have been put out on, on these models and I think it helps to read them. You don't have to read all of it, but I think it will be helpful for you to ground yourself in some of this research to understand kind of how AI has evolved and where things are going. Um, and so that, that would be the last piece of advice that I would have for you.

Um, and with that, I will, I'll close up with, with this final statement, which is this world is dynamic and it's really incumbent upon all of us to be as dynamic as the world that we live in today and tomorrow. And that means we need to continue to be adaptive, we need to be continuing to be comfortable being uncomfortable. Uh, and, and that's really the way I think forward.

Now for, for more information about me and myself, you can follow me online on LinkedIn under my name Colin S. com. I'm also on Twitter slash XC Levy Law, that's C-L-E-V-Y-L-A-W.

And if you're interested in learning more about the intersection of tech and law, uh, please check out my book on Amazon, the Legal Tech Ecosystem. Uh, it came out in October. I'm really excited about it and I think it helps ground you in this world of legal and technology and it's really an accessible introduction, uh, through some of the many conversations I've had with leaders in the space.

So I really appreciate you all listening. I hope this was somewhat helpful and gave you a taste for what's to come as well as what exists now. And thank you very much and have a great rest of your day.

I.