A Privacy Lawyer’s Guide to AI Implementation

Laura Jeffords Greenberg: We will get started. So thank you so much for joining us. Laura Jeffords Greenberg: For a privacy lawyer's Guide to AI Implementation. Both myself and Ben Martin are very excited to have you here, and we are just going to jump in and get started, if I can make sure that my computer is properly working. Laura Jeffords Greenberg: There we go. So since Ben appears first, I'll let him introduce himself, and then I'll introduce myself after. Ben Martin: Thanks, Laura. Hey, everyone. Yeah, so my name's Ben Martin. I'm Head of Privacy and also a legal engineer here at Wordsmith. I've recently joined, this is my… actually my third week, but formerly I was a director of privacy, at… Ben Martin: Trustpilot, so I was involved in the implementation of AI there, and I'm drawing on that and my new experience to talk about implementing it in today's session. Laura Jeffords Greenberg: Yes, awesome. And you, yeah, the use cases that you had were really great, so I'm excited to hear more about those. Laura Jeffords Greenberg: I am Laura Jeffords-Greenberg, and I am head of the Wordsmith Academy, which, since you are here. Laura Jeffords Greenberg: you have joined and you are aware of, and so my job is to educate the legal community about AI. My specialty is how do I use Laura Jeffords Greenberg: AI in practice, but I want the Academy to address everything AI-related, so we're trying to bring in more specialists like Ben to talk about different substantive areas Laura Jeffords Greenberg: Next week we're going to talk about, you know, how do you buy AI, illegal AI tools, so we're going to really try to cover everything, and I'm excited to get kicked off with privacy. Laura Jeffords Greenberg: And so, like we usually do, we're gonna start off with a couple of questions for you, because we like to start off with engagement. Laura Jeffords Greenberg: So what we're gonna do is I'm gonna actually jump over here to Slido. So you can either use the QR code, join Slido, or I'm just gonna look over here, because I can't… it's too tiny over here. You can go to Slido.com, and it's 2456377 is the number. Laura Jeffords Greenberg: And if you would go there and let us know how often do you use AI tools in your legal work? Is it hourly, daily, weekly, monthly, or never? Laura Jeffords Greenberg: Whoo, alright, we had someone, an answer, yay! So we'll give you a chance to do that. So we've got hourly, someone with hourly. We're just gonna get an idea. Every time we do this, we're trying to see, you know, who are we speaking to? Laura Jeffords Greenberg: to make sure that we can tailor our discussions and our talks to the audience, because people are, you know, in very different places in their AI journey, depending the industry that they're in, their role, their company policies, right? Sometimes companies may not be letting you use AI. Laura Jeffords Greenberg: At work, but we'll let you maybe do it in your personal life. Laura Jeffords Greenberg: I do have to say that I am seeing this increase. I ask this same question, with almost all of my presentations, and we're getting a lot more, Laura Jeffords Greenberg: Where people are using it hourly. Perfect, so I think hourly here is definitely the winner, but people are at least using this. Laura Jeffords Greenberg: Daily, weekly, hourly, so they're… we have AI users on the call, which is great. Laura Jeffords Greenberg: And then we have one more question. Laura Jeffords Greenberg: When you've tried AI tools, what best describes your results? And when forcing you, no, you know, it depends, lawyer answers, game changer, mixed results, or frustrated. So you kind of got to pick one. Laura Jeffords Greenberg: Again, this isn't scientific, but we just kind of want to get an idea of, you know, the group that we're speaking to. Are they… Laura Jeffords Greenberg: into AI tools, or are they disappointed, by them? Laura Jeffords Greenberg: Perfect. Okay, so we've got… we've… oh, split, sort of. We don't have anyone that's super angry or frustrated, which is probably good, but we have game changer and mixed results. Before I hand this back over to Ben, right? That's what we're doing? Laura Jeffords Greenberg: Please feel free to introduce yourself in the chat, feel free to throw your LinkedIn message, in our LinkedIn Laura Jeffords Greenberg: link in there. Sorry, it's been a long day. I'll stop babbling now and hand it over to Ben, who will inform us about all the awesome things he's done at the intersection of AI. Ben Martin: Yeah, thanks, Laura. So is everyone able to see Ben Martin: the main screen saying, what was my experience? Great. And it was really interesting seeing those different, statistics and where people… where people are coming in. It sounds like everyone on the call has some really good experience with AI in terms of they're using it, but there is some mixed experience around how useful that is, or how positive that… or how good the output is. Ben Martin: So… Ben Martin: Hopefully, with today's session, we'll be able to go through and help you on your journey, and give you some of my learned experience, which I hope you will find helpful. Ben Martin: Before I sort of run through what we're going to look at. Ben Martin: What I thought might be helpful would be to talk a little bit about my experience, just to set the scene, and the problems and the things that I was facing in my last role. Ben Martin: So… We had a really significant influx of questions and problems from customers and consumers and colleagues. Ben Martin: And our team was in a very reactive position, and we were unable to get to those really important areas of work, to be proactive, to do the strategic projects that we wanted to be… wanted to be doing. Ben Martin: So we were stuck in the day-to-day, and we were doing things in a very manual way. Ben Martin: To compound this, we also didn't have budget to add resource in terms of bringing new people in, despite an exponential growth in workload. Ben Martin: So, the team and I thought it would be good to Ben Martin: try and implement AI to help with each of these issues. So, not only the high-volume tasks, which would enable us to switch away from being proactive and work on those more, business-enabling projects. Ben Martin: But also, something which I think's really important where you have a team, is to make everyone's job more enjoyable, and help them work on things which will grow and develop them, which I don't think is always talked about in this sphere, but I think it's something that I'm… I think's really important. Ben Martin: So, to go back to those questions. Ben Martin: For me, it was something we went from using it a little bit, to using it every day, and then to every hour. Ben Martin: And the results were really transformative for the team. Ben Martin: So… that's kind of my experience, and how I'm going to break that down in the session is… Ben Martin: into these four areas. So… Ben Martin: We're going to start with looking at some use cases, because I always find these really interesting to see what other people are up to. Some of these I've used myself, and some are new ones that we're working on with Wordsmith that I'm quite excited about. Ben Martin: Then we'll move on to key learnings. So, starting off with preparation and helping you prepare for success. Ben Martin: Before moving on to getting started, like, how to implement Ben Martin: and then looking at some challenges. So… Ben Martin: some of you seem to already be quite a long way in your journey, but I hope there will be some Ben Martin: Things that can help you in terms of the implementation, and the different angles and viewpoints to what you've been looking at already. Ben Martin: And what I would say is I'll pause in some places for questions, so please feel free to just ask as you go, come off mute. And Laura, if you spot anyone putting their hand up or shouting out that I miss, please give me a nudge. Ben Martin: And, we'll also have some time at the end to run through as well. Ben Martin: Cool. Any questions so far? Ben Martin: I'm gonna take that as a no. Good. Okay, so we're gonna look at some that, as I said, that excite me. Ben Martin: and some that I've used before, but something that I found, certainly within… when I started using Wordsmith, was… Ben Martin: utilizing the legal layer AI, so a legal-specific AI, that was a really helpful thing to play around with on research tasks and general chat in the same way you might use another LLM. So, putting that to one side, I think that's a really good use case, but I wanted to just jump into some slightly more Ben Martin: exciting ones. So… This first one is… Ben Martin: We'll hit play on the video, and… Ben Martin: I thought, rather than doing a live demo, it'd be better to do a recorded video and talk to it. Ben Martin: What I'm going to show you now is effectively a chatbot, and it's going to use privacy documentation to… Ben Martin: answer questions. And it's been built in… on Wordsmith, what we call workspaces, so I'll just show the navigation to that. I've already set it up. And you can see here that you've got Ben Martin: the knowledge… on the right-hand side. On the left-hand side, you've got a familiar-looking chat box. Ben Martin: And you can see I've already put one document in there. Ben Martin: So, I'm just gonna add an extra few documents on the recording. Ben Martin: And they will load up into the knowledge base. Ben Martin: And what this means is that when you ask a question to the chat, it will look at that specific knowledge. Ben Martin: and draw the answers from there. So… Ben Martin: the benefit of that is that you can have a specific topic, be it a legal topic, finance, privacy, where you feed the information into it, and it will populate it. And I'll show you some extension strategies, which I think are really useful for using that in a business in a moment. Ben Martin: So if I hit play again, you can see that, as well, you can give it some custom instructions. So. Ben Martin: you can have it just answer questions as the AI would normally, but I've just thrown in some very basic instructions on a stylistic point of view here, just to make the output useful for the recipients. Ben Martin: We'll just go back to the… the interface. Ben Martin: And then… Based on the information we've got here, I'm just asking a simple question. Ben Martin: About data retention. Ben Martin: And… Ben Martin: the Wordsmiths AI will generate an answer to that. Now, this is obviously useful for me if I want to quickly look up data retention information, but where it can be really useful is where you link it to… if you use an instant messenger like Slack, you can link it to Slack. Ben Martin: very, very easily, and it can answer questions in Slack, which will allow people to self-serve within your business. So, the way I think about this is… Ben Martin: It's… Ben Martin: like an assistant who can answer all of those annoying little questions that distract you and pull you away from your work throughout the day. So, you know, who's the company's signatory? Where is our data stored? Who processes our data? Ben Martin: That kind of thing. And it can give people the answers that they need to do their job without disturbing you. And finally, you'll just see here, as well, you can reference the documents, so it will cite the documents and pull that in, which I think is a really neat feature. Ben Martin: And… Ben Martin: when we implemented a similar tool to this in my last role, it saved us up to 50% of the time we were spending on those same questions. And I'll talk a little bit more about how we came to that data later on. Ben Martin: If you're interested in this, there's a QR code which you can scan in the bottom right, bottom left, sorry, which will take you to a detailed view of how to actually implement the tool. So, I'll pause there in case any questions Ben Martin: Cool, okay. Ben Martin: So, the next use case I want to look at Ben Martin: is horizon scanning. Now, I'm a privacy lawyer, so part of my job is looking at new legislation that's coming down, new guidance, new laws. Ben Martin: what's going on with the market, and what competitors are doing. And… Ben Martin: you can use this tool to scan the horizon and really get an idea of what's going on, and it's really customizable. So, what you can see in the top left is a source that I've pulled into… we use Slack here, so I've pulled it into Slack. Ben Martin: And that is just a summary. So that's using what's called an RSS feed, and that pulls it in as a summary. Ben Martin: But where it's really useful to add Wordsmith in is you can create a specific summary, pulling out the points that you want, and stylizing it how it's useful to you and your team. Ben Martin: So, for me, it helps digest a really massive amount of information and makes it very… it synthesizes it and makes it very easy for me to take a look at. So, it's not something that is… Ben Martin: like, not only something to free up your time with AI, it can also enable you to, like, focus on things you're interested in, or, for me, it was supplementing my knowledge and keeping me informed of what's going on. Ben Martin: And my colleague Ellie has done a really good video on how to build one of these, which you can build the first part Ben Martin: even if you don't have Wordsmith, and there's a guide to that if you follow that QR code there. Ben Martin: I really like that one. Ben Martin: Okay, finally, this is a new tool that we are in the process of releasing at the minute. And so, this one I thought was really cool, and it's effectively a… what's been called… we're calling it Blueprints, and they are intelligent forms Ben Martin: Which can draw on multiple sources of information to guide Ben Martin: Wordsmith's AI to fill in and complete impact assessments, COSEC forms, and other documents. Ben Martin: So, here we can see a data protection impact assessment, which, if you've ever done any privacy work, I'm sure you'll be familiar with. And this is based… this is based on the ICO's, Information Commissioner's Office template. Ben Martin: So, what we've got is this is all preloaded in, it's got wordsmiths. Ben Martin: the legal layer of AI within Wordsmith, and then we've built… my colleague Emily has built a really good, Ben Martin: Legal intelligence on top of that to specifically fill in this form. Ben Martin: And… Effectively, you can bring in 00:17:29.670 --> 00:1746.060 Ben Martin: your information that you would use to complete a DPIA. Typically, you're going to have the data processing agreement, you might have a privacy white paper, security information, and you can even bring in jurisdiction-specific guidance. So, if it's a UK Ben Martin: business, you could bring in UK guidance. If it's European, you could bring in the EDPB guidance. Ben Martin: And… you can… Get it to… to run the… Ben Martin: the DPIA and complete the DPIA in the background. Ben Martin: So… I think… Ben Martin: we're doing a very light-touch one here, just with the DPA for… to demonstrate it, but this… you'll see how quickly it does it. And I… I think this is really useful, because it… Ben Martin: even if you were just using it to put the DPA in, it's very, very quick, and it brings a lot of the information that would take a lot of time to write into the DPIA. Ben Martin: Where it's more powerful is if you can bring in that other context, that other information, put your call notes in there, and then that can build out quite a detailed DPIA, which has Ben Martin: Saved you an enormous amount of time, and again, has… is well referenced, is… Ben Martin: having… pulling from… strands from all of the different documents. And I think, of course, with Ben Martin: an impact assessment, if you're doing this, or… there's also going to be transfer risk assessments. Of course, if you're doing these assessments, you want to have someone reviewing it, either DPO, a lawyer, but what it does is it does that first Ben Martin: draft of it, and it does it to a high standard from the testing that I've done. Ben Martin: And because it's a legal-specific tool, it can get you a lot of the way there. Ben Martin: So, you'll see it's pulled out the draft from the DPIA, it's also referencing each of the points and where you can find the information, and as I mentioned, it becomes more detailed the more information you put into it. So, I think… Ben Martin: Personally, this is going to be a game changer, because it will mean that it becomes much easier to not only do the DPIA, Ben Martin: But check that DPIA on an ongoing basis. And also, because it's quite a quick process. Ben Martin: it will mean that where you perhaps didn't always have the time to get round to doing that DPIA, it will mean you can do one, file it away, and you've got your thinking Ben Martin: They're written down. Ben Martin: And yeah, so that's just finalizing it, and you can see there the… Ben Martin: It's completed it with my name because of the, the profile, so… Ben Martin: Yeah, so those are the three use cases I wanted to chat through. Ben Martin: If there aren't any questions, I'll continue, but I'll just pause a second to give anyone a chance to ask any. Laura Jeffords Greenberg: I… can I… can I ask a question? Ben Martin: Yeah, of course. Laura Jeffords Greenberg: Is… I mean, we're showing this, obviously, in Word… Wordsmith, but in theory, we could do similar things in other tools as well, using the framework or the thought Laura Jeffords Greenberg: Logical thought process of putting this together. Laura Jeffords Greenberg: Do you think that's possible? Ben Martin: So, I… before I came across this, or before I was working on this, again, in my old role, I was using… Ben Martin: I would use LLMs to help with DPIAs. Ben Martin: what I think the difference is, is firstly, it's a secure environment, so it's not… you're not, kind of, putting your… Ben Martin: put sensitive information, potentially personal data, commercially confidential information out there into the ether. So, it… for me as a lawyer, it reduces my anxiety level, which I think is always a good thing. And second, there's the… Ben Martin: not only the format, which is really nice, and Emily's done a great job making it look so good, I think, Ben Martin: But the… the way it goes about solving the problems Ben Martin: the, like, Wordsmiths AI is, which I'm still learning a lot about, The output is more… targeted for… Ben Martin: like, legal, and specifically DPIAs in this case. So, the checks that we've built Ben Martin: for the DPIA specifically are really, really help it give the right answer that you want, and target the… the questions, in the DPIA. Laura Jeffords Greenberg: So I guess, yeah, it's a combination of… Laura Jeffords Greenberg: that legal intelligence, right, that we've put in there, and then the workflows and integration with Word. Laura Jeffords Greenberg: because I love ChatGPT, obviously I use WordSmat, too, for legal work, but, so far the other general LLMs, like, don't work with Word in the same way. Laura Jeffords Greenberg: Yeah. That wordsmith does, and so I was just wondering what else, and so there's that privacy layer, and then a workflow layer, so there's kind of three differences between… Laura Jeffords Greenberg: the large language models, and that we've built that in so it's easier. And again, this is not supposed to be salesy, I'm just… I always like to say, can we extrapolate, like, what we've learned here and apply it with general? Laura Jeffords Greenberg: Broad-purpose tools, but it seems to be more of a challenge that way. Ben Martin: Yeah, I think there is… the way I… when I was doing it more generally, I was having to just basically try and strip out anything which would identify the companies, anything personal, when I was using it to write the DPIAs, and obviously using… Ben Martin: modes where it was not training… used for training data and things like… Training models and things like that. Ben Martin: So it is… it is definitely possible. I… from my experience, I think this is a superior thing, and I'm obviously… I'm not saying that because I work here, I just… I think it's great. Emily's done a great job building it. Ben Martin: But yeah. Ben Martin: Okay. Ben Martin: So… onto my learnings from implementation. So this is going to be really practical. Ben Martin: And… I will… Go through before, during, and then some of the challenges. Ben Martin: So… Starting off with… Ben Martin: AI implementation equals change management. So, my view is that if you want AI implementation to be a success, you must view it as a cultural change within your team, and potentially your whole organization. Ben Martin: Because Ben Martin: it will change how you interact with stakeholders, and although there is a technical element, this… I think it's very… it's very much a human-centric change. Ben Martin: And… Ben Martin: within… if you are someone who wants to implement this in your team, or you're excited by it, I think the whole… it's really important to bring the whole team along on the journey. So, like with a treadmill. Ben Martin: you will only get fit if you use the treadmill, and are committed to doing so over a period of time. I think buying an AI tool is not going to solve the problem. Ben Martin: There's also, I think, the importance of… Ben Martin: not thinking that AI is a silver bullet that can do everything with no input. So there's work that's required to learn how to use the tool, and it can cause growing pains within a team, especially if there are embedded processes where someone has been doing something a certain way for an extended period of time. So there's a real adjustment period. Ben Martin: As well as within your team, I think you need to bring others in the wider business, those people who are on your workflow management tool if you use them, or teams who interact with you through Slack or email. So they need to be kind of brought in, you know, give them a hug and bring them into the circle as well. Ben Martin: So… Ben Martin: I think the important thing there is that it's a change in behavioral management that takes time and effort in the short term, but longer term, I think there can be benefits. Ben Martin: The next point I've got there is finding a junior or mid-level AI champion. Ben Martin: And… I think, fortunately, AI is quite an exciting topic. I mean, there's people joining the call at… Ben Martin: in Europe at 5, 6 o'clock at night, so there obviously is interest here. And it can be easier to get buy-in from junior members in the team. Ben Martin: Now, I think it's really important to have a junior or a mid-level person as your AI champion, because they're the ones who will be much closer to the day-to-day, of implementing the tool. So, they're the ones who are going to be knowing how the workflows work, and Ben Martin: when you're looking for someone in your team to be the AI champion, I think you want someone who's really genuinely curious about it. Ben Martin: has a growth mindset. They don't necessarily need to be super technical, but if they're excited and interested about learning about AI, that's really beneficial. Ben Martin: They've got a good instinct or experience for finding pain points and looking at processes, and also someone who can Ben Martin: Influence others and is trusted by stakeholders in the business. Ben Martin: So that's getting buy-in from within your team. Ben Martin: I think the other thing that obviously is really important here is getting Ben Martin: stakeholder… Sorry, getting senior buy-in, because… Ben Martin: this needs to be done from the start, and I think you can do this by trying to come up with potential benefits for you and your team, and taking them to leadership and presenting them along with any data you can get from vendors. Ben Martin: A high-level plan to clearly articulate where you are now, and the problems, and where you would like to go, and the things that you want to solve, and being really, kind of, specific with those. Ben Martin: Fortunately, we had someone Ben Martin: further up the chain, I was very keen to try it, but I think if there were issues, you could perhaps try to go down the route of, can we do a trial for 1, 3, 6 months, rather than just saying, let's sign up to this new tool. Ben Martin: And I think all the usual things apply here, of when you're Ben Martin: Trying to get people on board, and when you're doing a trial, over-communicating, having regular catch-ups, producing ongoing data, and explaining the benefits in terms of not only the time and cost saving, but also morale, development, and all those other softer pieces as well. Ben Martin: So I think the preparation side, drawing this together, is about people. It's about getting the team on board, communicating and getting stakeholders involved, having a champion, and then getting senior leadership buy-in as well. Ben Martin: Once you've… Got your plan together. Ben Martin: And you've perhaps got that buy-in, you've found some vendors. Ben Martin: You want to start looking at who to select and who to work with. Ben Martin: And what we did in the team was to run a workshop to generate ideas and potential use cases for the AI. So, this is… Ben Martin: getting people to dream big, come up with as many ideas as possible, and use it to generate some excitement. I think, again, it's that kind of giving people space to play and have fun with it, rather than Ben Martin: Trying to be too, Ben Martin: Too fixed as what you might have from the start, just letting people come up with ideas in the workshop, and then refining it. Ben Martin: So… Ben Martin: I think that's… that's really important, is creating… gathering ideas from the team, getting people involved. And then, also, this can be a really good opportunity to help identify potential AI champions who can then help you with the vendor selection. Ben Martin: I know Laura's doing a session on vendor selection, so I'm not going to touch on that now, but that's happening next week, so I'll skip over that, but… Ben Martin: You can learn more on that next week. Ben Martin: Next, I think you need to create space for your AI champion. So, I think there's a real risk that it just becomes that other thing that someone has to do, unless they have the space to work on it. Ben Martin: And it's best not to just say, okay, here's an extra task you've got to do, because often people don't want to take on more work if they've got a full plate. So. Ben Martin: what we did is we tried to carve out space for our AI champion, and she was excellent and really took it on, and… Ben Martin: kind of was one of the key reasons the whole project was a success. Ben Martin: So, obviously, we've been talking about getting your team on board, so it's like, can they potentially help on… in the short term? Can they pick up some of the slack? Ben Martin: And having that time-banded 1, 3, 6-month period to… to just, you know, come together as a team. Ben Martin: Also, if you're doing a trial. Ben Martin: And you're within a legal team, so for us it was a privacy team. Ben Martin: We… it was being viewed as a pilot for the rest of the team, so we were able to get some extra resource and help from other areas of the legal team, but we also spoke to Ben Martin: Support and other areas like that to bring in Ben Martin: resource, even just to help pick up a little bit of the slack. So that… that is another area I think you can create a bit of, bit of space for your AI champion. Ben Martin: Finally, I think the… Ben Martin: the potential to build it in as, like, a key objective for your team, so if you have OKRs, objectives, and key results. Ben Martin: Building the success of the project or building the trial into that can help not only get visibility, but also get it, like, on the agenda and make people focus on it. Ben Martin: as you're starting to move into getting things running, I think starting small with easy wins is a really good way to do… to do this. Doesn't need to be glamorous. Ben Martin: But it can be a problem where there's… Ben Martin: AI could genuinely help, so it's something real and specific that is causing pain. So, you could take that Q&A bot, for example, and come up with a first version that you Ben Martin: have, and you use it to ask it questions. The next step could be bringing it into one of your workflows, so bringing it into Slack or something like that, so you can open it up to Ben Martin: people in your business, or you can still verify it before it goes to them, but having that… people being able to self-serve, and then further down the line, it could be a… Ben Martin: something on your website that your customers can go to directly and ask it questions. Obviously, you need to do a lot of checks in each of those scenarios, but Ben Martin: Testing the tool, getting the feedback, and then iterating is a really good way to just slowly build up the tool, and to demonstrate what it can do. Ben Martin: when you're… thinking about implementing a tool. So, let's say you're building a contract negotiation tool. Ben Martin: You need to ensure that your templates and playbooks are in good order, and similarly, if you were doing the Ben Martin: the Ask Privacy type, Q&A bot. Ben Martin: it's important to ensure that you've got comprehensive information that's really good quality and is well sanitized as well. So. Ben Martin: that whole thing of, like, if you put something good into the AI, it's more likely that something good will come out. I think that is really important, and it ties back to what I was saying at the start about not thinking it's a silver bullet that will do any… everything without putting any work in. Ben Martin: Finally, and I think this is… this is something that, we had… we found really useful, was gathering data on the impact, even if it's simplistic. So… Ben Martin: Obviously, when you're starting out, you haven't got necessarily a full change team, and project managers, and everyone doing metrics, and a data scientist, and everything else, so… Ben Martin: you might have to just do rough estimates. It might be, how long are you spending without AI per day on this area of work? Ben Martin: That is then automated, and then you ask someone. Ben Martin: How long, roughly, are you spending now? Ben Martin: Obviously, that's not an ideal situation, so you could take a… still a low effort, but an improvement there, and there's free time recording tools that you can use online, so perhaps doing a period of time before you implement the tool of, like, a month, or two weeks, or whatever you feel is right. Ben Martin: And then one went shortly after you've implemented it, and then periodically throughout your trial period. Ben Martin: To see the changes in the areas of work. Ben Martin: And if you map out all the areas of work that your team are using AI for, and you can compare the before and after, I think that's a really good way to not only help you understand, but also help others understand. Ben Martin: Which takes me on to my next point, which is… How to deal with challenges. Ben Martin: So… Ben Martin: I'll talk to the data point in number 3 there, but the challenge is, you know, you say, oh, this all sounds good, Ben, but, like, ultimately, life still goes on. We've still got the workload, we've still got increasing… Ben Martin: Increasing pressure, and it's hard for someone to carve out the time. Ben Martin: So… Ben Martin: I think the… my response to the challenge… that challenge, is it has to be a short-term versus long-term thing. So, in the short term, you might be doing… Ben Martin: it might be a short-term pain, almost, if you have to potentially lose some capacity from someone in your team, but long-term, the results can be really beneficial. As I mentioned before, we were saving 50%. It was taking us 50% of the time on some of the work that we were doing. So, absolutely huge time savings. Ben Martin: Next is, okay, well, Ben, you saved 50% of that time. What are you doing with it, with all the free time you've now got? Now. Ben Martin: Again, I think this comes down to having data. So if you can… and in my experience, the team were over capacity prior to us implementing AI, and that was only getting worse and worse because of the increasing workload. What it did was it helped to bring the capacity down. Ben Martin: Sorry, bring the workload down, and… Ben Martin: give people more capacity that then enabled us, A, to not feel like we were underwater as much, and B, it enabled us to start focusing on more strategic matters, that we could start doing the Ben Martin: doing regular audits, doing all these things that we wanted to be doing, but it was always the thing that we never quite got to. So… Ben Martin: it enables you to be proactive, and I think Ben Martin: Having a… perhaps having a plan to deal with that question, as and when it comes, is really good to do, and also having the data to back it up. Ben Martin: Helps you spin the narrative. Ben Martin: So, finally, just on that data point, you will have… you know, it sounds like there aren't skeptics on this call, but there might be skeptics out there, so I think the… the thing that I'd like to just say is, like, when you come up against skepticism and inertia, if you are… Ben Martin: keen and want to implement AI within your business. Ben Martin: the… my approaches for dealing with that are, first of all, always asking for a trial. I think people are open to a trial because they can see it. Okay, well, if it doesn't work. Ben Martin: We can go back, there's no issue. Ben Martin: Next is that data point, having data to support Ben Martin: Skepticism, and if you can get data potentially off of vendors, that can help with the inertia at the start of the project. Ben Martin: Also, I think… Ben Martin: talking about well-being, morale, is it more fulfilling, is it helping development within the team? That, for me, is important to me, and… Ben Martin: you know, whenever I've managed people, I always want that to be at the top of their… like, them to have… top of their mind, them to have capacity to learn and develop, so I think that's a really important thing. Ben Martin: perhaps it won't necessarily always convert the skeptics, but it's an important one to throw in there. And also, like, showing cool use cases. I think I showed you those ones that I found exciting and interesting. Ben Martin: they might have struck a chord with you, you might also say, well, that means nothing to me, it doesn't help me negotiate contracts quicker, but I think if you find the right examples, you can really get people on board, and Ben Martin: That's a way to get people excited by it. Ben Martin: And I think I've not put that up there, but finally is just… Ben Martin: creating an environment where people feel comfortable asking questions. I know when I started. Ben Martin: learning about AI, and I still consider myself very much in that camp of learning about it, but when I started learning about it, it was… it felt intimidating, because, oh, other people in the room know more than me, or… Ben Martin: I don't want to look stupid here. As a lawyer, I'm always supposed to know the answer, right? So, I think creating an environment where people feel comfortable and it's a friendly atmosphere is a really good way to get people on board. Ben Martin: Okay, last one, before we have some question time. I just wanted to round up Ben Martin: hopefully I've been hammering home my thoughts across these areas throughout the session, but just to summarize them. So, I think thinking about Ben Martin: AI implementation as a people and change management project. That's my first point. The second one is having someone enthusiastic, championing Ben Martin: this project in your organization. And the third one is to use data and concrete benefits to demonstrate the efficacy of the project and the benefits to your team. Ben Martin: And that's me. So… open the floor to questions. Feel free to put them in the chat, or… Ben Martin: Yeah, we'll just shout out. Laura Jeffords Greenberg: There, there are two questions in the chat. Ben Martin: Okay. First one was mine. Laura Jeffords Greenberg: So I'm gonna go first. What were the OKRs that you used around AI adoption? Because I know that I've talked to a lot of leaders about this, and so what… what were some of those? Ben Martin: Yeah, so… Once we had as I said, we were… had a supportive Ben Martin: Chain of command in terms of… Ben Martin: getting AI in the first place. Ben Martin: But when we… once we'd done an initial trial for a couple of months, we started to set out goals for reducing the workload time, so it would be Ben Martin: OKR for the first quarter was knocking it down by… I can't remember what it was, but let's say it was 15% less times. Ben Martin: yeah, 15% less time spent on the tasks, and then a lower one, and then a lower one. So that was one way we tried to do it, to focus it on timing, which meant that eventually you had to come up with better and better use cases to save more and more time. Ben Martin: The other option was implement… once we had tested something, or proved that it was a good beneficial… tool. Ben Martin: It was like, okay, so… Ben Martin: that's… we've done a test of that. Now, the goal for the OKR is to have that fully rolled out across these other teams, or fully rolled out Ben Martin: With these extra features built. So it's tied to either time-saving or features As part of the tool. Laura Jeffords Greenberg: And did you have usage at all as OKRs for individuals? Laura Jeffords Greenberg: We've seen that at some companies, and I don't know whether that is the right approach. Ben Martin: Yeah, so we… we had our AI champion who was owning those goals, but I think the… Ben Martin: the culture that we had within our team was not, like, if something wasn't achieved, it was more of a, okay, well, that hasn't been achieved, why is that? Let's fix it and enable you to achieve it. And I think that comes back to my, like. Ben Martin: I guess it's, like, the culture point, which, you know, I could do a session on what I think on culture, like, another time if you want, but, I think, yeah, having the right culture makes it more likely to be implemented. Laura Jeffords Greenberg: And I think John had both a question in the chat and raised his hand, so I think he's up next. Ben Martin: Hey, John. John Cameron: Yeah, I may have… thanks, Laura. Thanks, Ben. I may have, John Cameron: Variations on the same question, but, John Cameron: I think Laura probably knows this. I mean, I'm an absolute huge fan of, John Cameron: using AI, and I'm even more a fan since I hooked up with Wordsmith, so you can take it John Cameron: I get it. John Cameron: I was on a… Well, I'll start with my question in the written chat, which is, John Cameron: Is this… if you can show metrics and data John Cameron: showing the improvements in efficiency for a legal team in a corporate environment through using Wordsmith or another AI tool. John Cameron: Isn't that… John Cameron: in the real world, effectively, talking senior lawyers out of jobs, because the CFO is going to be looking for John Cameron: efficiency gains and reducing costs. So, if you convince them that you can do a lot more. John Cameron: with fewer staff using AI, you know, is it not effectively a race to the bottom for senior legal professionals? John Cameron: And… as an aside to that, I was in a… Group chat on, the Crafty Council Consultants Group. John Cameron: Today, about the fact that there's… Fewer and fewer… Senior-level legal jobs being advertised. John Cameron: by scale-up companies, and I think a big part of that is I think… John Cameron: scale-ups in the UK environment, I think it's very different in the USA, but UK scale-ups don't really budget for senior legal hires, and there's increasingly a feel that you can hire a lawyer John Cameron: possibly part-time, and rely on… AI to fill in the gaps, and, you know, that's legal ticked off. John Cameron: So… I mean, for me, there's a general feeling that… John Cameron: It's great for producing high-quality work. John Cameron: at pace, but I think it's terrible for the practitioners John Cameron: And my last point on that would be a personal one as a… John Cameron: fractional general counsel, I can see enormous efficiency gains. John Cameron: and quality gains through using Wordsmith and other AI tools, so it's a boon for me personally, but… John Cameron: Much as everybody says the hourly rate's a terrible thing and a bad idea. John Cameron: Most of my clients haven't got the memo on that, and they want to be charged an hourly rate. So, effectively, if I told them the benefits I'm getting with AI, John Cameron: And gave them an actual time count that would slash my income at the same time as I've got much increased expenses paying for the tools. John Cameron: I think they're all variations on the same point. Ben Martin: Yeah, yeah, loads of, loads of really interesting points there, and good, good challenging questions, thank you. Ben Martin: Can I just ask, you mentioned about the headcount piece, and you said senior lawyers. I find that interesting, because I often think that the argument is more that there will be fewer juniors, but what was your thinking there about the senior side? John Cameron: I think it is because, what I and others see in the market is there's fewer and fewer senior hires. I mean, I mean, there's an economic point that everybody's nervous, and… John Cameron: People don't know how the economy's going, and NI's gone up, but I come from a regulated fintech background, and John Cameron: Whereas… I'm checking what year it is now, I mean. John Cameron: 12 years ago, I was hired as a fintech GC, going into a, you know, regulated environment in a startup. John Cameron: And… I was taken on on a fairly full salary. John Cameron: And that was just seen as the right thing to do. And increasingly. John Cameron: You know, the regulation hasn't reduced since then, but increasingly fintech startups and scale-ups our recruiting… John Cameron: for one or two year qualified lawyers, and aren't hiring the senior people, and I think AI probably just, John Cameron: magnifies that trend. Ben Martin: Hmm. John Cameron: That seems a cost as a senior person. Ben Martin: Yeah. John Cameron: I mean, I agree entirely that there's going to be an impact on junior people as well, because they might not get to first base, because you just think, well, I don't need people doing due diligence, I don't need people doing contract reviews, I just need somebody to look at the outputs and know what they're doing. Ben Martin: But… John Cameron: I think a lot of… John Cameron: founders and CFOs see lawyers as being pretty fungible. It's like, you've got a law degree, you're a lawyer, so I'd rather have the cheap one who's just got off the plane from Australia and is 2 years qualified, rather than the 20 years experience guy. Ben Martin: Yeah. John Cameron: old lady. Ben Martin: Yeah, no, that's super helpful, thank you. I wonder if… there's quite a few things that I'd like to respond to there. I think… Ben Martin: In my experience, and this is not… leaving aside the senior point for now, but in my experience in my last role. Ben Martin: we were… it wasn't a case of the reduc… it was, like, reduction of headcount, because if… let's assume… let's say we were at 130% capacity within the team. That… that seemed… I think that's probably what it was. Ben Martin: If we used Wordsmith. Ben Martin: for a number of areas, that would bring the capacity down. It wouldn't bring it down to 20%, because there's only certain areas that you can use it, you know, it's the more… we were using it for question answering, the more Ben Martin: routine… time-consuming stuff. Ben Martin: So, for us, it brought the capacity down to a more manageable level. Ben Martin: And it enabled us then to work on Ben Martin: Projects to improve things, working on strategic projects. Ben Martin: doing the things that we wanted to be doing, but were either too busy to do, or didn't have the time. And… Ben Martin: What it became was a tool to redeploy people onto other things, rather than a tool to say, okay, there's no… Ben Martin: there's no role here. And for the organization I was in as well, it was a rapidly growing organization, so the workload was growing, and if we hadn't brought it in, I think we would have struggled to actually maintain the level of service that we were providing. Ben Martin: So then, your second point was around the, like, hourly rate side of things, and… Ben Martin: Yeah, I've been seeing more and more articles now Ben Martin: how AI is challenging the law… certainly the law firm model, consultant model, on hourly rates, and… Ben Martin: I don't know if I'm the best person to speak to that, because I've spent most of my time in-house, but I think… Ben Martin: It almost… it does ask a very big question of, like, how will consultants and… Law firms Ben Martin: start to change their billing. I don't know, it'd be interested to hear if you have some thoughts on that, actually. John Cameron: Well, I mean, in an ideal world, I would, John Cameron: Quote, a price for a successful outcome on the job. Ben Martin: Yeah. And it's not… John Cameron: the client's concern, how long I spend on it, as long as I don't take an unduly long time. Ben Martin: But as I was saying. John Cameron: I've been doing the fractional GC thing for 2 years, and John Cameron: I struggle to get most of my clients off an hourly rate. Ben Martin: Right. John Cameron: I really struggled. John Cameron: But, over the same period. John Cameron: my outgoings on, you know, LexisNexis AI and on Wordsmith, and on ChatGPT and Claude, you know, have gone from John Cameron: Zero to a reasonably significant four-figure sum. Ben Martin: Yeah. Ben Martin: Yeah, I mean, I… I think there's a… there's a… there's still an open question on what… what AI does for… for pricing models, I think, and yeah, I… I don't have a… I don't think I've got the right answer to all the good answers. John Cameron: I mean, I'm not… I'm not expecting. Ben Martin: Me too. John Cameron: I mean, I may be letting off steam, but I think it's a fantastic tool, but it's a double-edged sword for a lot of practitioners. Ben Martin: Yeah, I would just… just one thought I had on the senior level piece, because, you know, I've recently been looking for jobs as well. Before I joined Wordsmith, I was looking for jobs at the sort of more senior end. Ben Martin: And it was very much… there weren't a huge number of jobs out there, and Ben Martin: whether that's due to AI or economic stuff, I'm not sure. Perhaps, as you say, in the smaller… smaller organizations which haven't got that budget, it might be a case where they just say, okay, we'll get the cheap person in because they're more junior, they'll be cheaper, great. Ben Martin: But I think the broader thing that I think about with all of this, the AI side of things is, if you have people in the business now who, with a bit of Naus and access to chat GPT, you can have someone in a product team or a tech team quickly Google the law and come to you with. Ben Martin: something that's… 60% of the way there, probably, 70% maybe higher on ChatGPT, if it's a general piece. Ben Martin: I think it starts to switch… certainly, again, this is more for in-house lawyers, but it starts to switch the role of being much more Ben Martin: A strategic thinker, and coming up with, like, solutions to help people Ben Martin: get to the right place, rather than want someone who knows the law. So I think you… at a senior level, your role will switch from being the person who knows the law Ben Martin: you know, this may be 10 years down the line or something, but knows the law inside out to someone who is much better at strategically advising the business. And it then comes back to people's skills and softer skills and that side of things. Ben Martin: That's, yeah, that's my view. John Cameron: Thank you. Ben Martin: Cool. Thanks, John. Laura Jeffords Greenberg: And I see we have about 5 minutes left, so, are there any other questions from the group that has joined us this evening? Laura Jeffords Greenberg: Very quiet. Laura Jeffords Greenberg: Are there, I guess, any tips or recommendations for individuals who, you know, want to learn more about Laura Jeffords Greenberg: privacy and AI, you know, anything that sort of relates to what you were talking about today, any recommendations of people to follow on LinkedIn or books to read? Ben Martin: Yeah. Laura Jeffords Greenberg: Wow. Ben Martin: Maybe I'll… shall I stop sharing screen as well, so we can get a bit more view? Yeah, so I think… Ben Martin: I think there's starting to be some really good… podcasts and… Ben Martin: regular updates out there. On the legal tech side, not necessarily AI-focused, but legal tech more generally, there's a podcast, Ben Martin: Which is run by… Ben Martin: two guys called Alex and Tom, and it's called Law What's Next? I don't know if people have seen that one, but it's very tech… Ben Martin: it's obviously law, what's next, what's going on in law. So I think that would be a really good, piece. I'm gonna have to get a plug-in for my book as well. I wrote a book about, that might have been a tee-up, Laura, so thank you. What! Ben Martin: yeah, so my book called GDPR for Startups and Scale-Ups, so it's basically making GDPR, accessible to people who aren't privacy lawyers. Ben Martin: So I'd recommend giving that one a read as well. Ben Martin: Hey, John, just in your question. Laura Jeffords Greenberg: Trying to get the link. John Cameron: Yeah, put my hand up as well, it's the same question twice again, but cool. Yeah, I got a data sharing agreement from a client. John Cameron: This afternoon, and they've got it from a business they are collaborating with. I mean, I've been using Wordsmith probably for about 6 months now, but, I'd appreciate your thoughts on them. John Cameron: how I best think about that, approach it, using Wordsmith as it is currently configured. Ben Martin: Yeah, okay, good question. I'll have to think of my feet on this one. So, data sharing, is that, like, a controller-to-controller? John Cameron: Yes, indeed. Ben Martin: Okay, who… Is your client the… Ben Martin: Import or the exporter in that scenario? Ben Martin: The sharer and the receiver. John Cameron: My client… is a manufacturer and… John Cameron: It is dealing with a… effectively, like, a sort of delivery service, which, delivers its products to… John Cameron: Retail consumers, or to places where retail consumers can pick up. John Cameron: The product that they have bought, so… John Cameron: by ICO guidance, the delivery company, you know, would normally be thought of as a joint controller. Ben Martin: Okay, so… I think… I would… Ben Martin: I would start by adding the… obviously adding the, Ben Martin: let's say we were doing this in chat, just to keep it simple. Adding it into the chat, uploading the file, and Ben Martin: giving a background about the scenario, and the key things that you are concerned about, from your perspective, and giving it as much context on that side of things as possible. So, let's say you had… Ben Martin: you put in your thoughts, but you also had a call note which explained that. I might bring that in there as well. Ben Martin: And… Talk about the concerns, the… the risks that you had identified elsewhere, and the things that you were… Ben Martin: yeah, I guess, like I said, the things that you were concerned about, and ask it to focus its attention on that when doing the red line. Ben Martin: And I think the other thing would be to… Ask it to… Ben Martin: obviously, I always like to ask AIs to ask me questions about, like, what more information do you need to know for this review to do it. Ben Martin: for my client. Ben Martin: And… Oh, I had one more point I was gonna say. Ben Martin: No, sorry, it's gone. But yeah, I think, I think the context… Ben Martin: Prompting the AI to ask you questions. Ben Martin: And giving it specific information on the things that you are particularly concerned about would be… would be 3 tips I'd give you. John Cameron: Perfect, thank you. I didn't know I was going to be reviewing a data sharing agreement when I put this in my calendar, but it just, popped into my inbox when I was on the train this afternoon. Ben Martin: Yeah, well, feel free to drop me a line on LinkedIn, and we can have a look at it if you have some feedback either way, good or bad. John Cameron: Perfect, thank you. Laura Jeffords Greenberg: Well, we are right at time. Excellent. Thank you so much for joining us, Ben, and sharing. I've put a podcast recommendation and also a link to your book Laura Jeffords Greenberg: in the chat, so that is there. This is recorded, we'll upload it to the Academy, so if you want to go back and watch anything, you can do that. And then, Ben and I both have very easy email addresses to remember. He is ben at wordsmith.ai, I am Laura. Laura Jeffords Greenberg: at wordsmith.ai, so you can always feel free to reach out to us there, or on LinkedIn. And, yeah, we love talking about AI, so please… Laura Jeffords Greenberg: you know, keep the conversation going, and also, if you could, tell your friends about the Academy, or do a LinkedIn post. We're trying to really build a community, and you're in on the ground floor, so we appreciate your support. There we go. And we have our LinkedIn right there, if you want to connect with us. So thank you so much, I really appreciate it. John Cameron: Thank you very much, both. Ben Martin: Thanks, bye. Laura Jeffords Greenberg: Good evening. John Cameron: Cheers, Ben. Cheers, Laura. Laura Jeffords Greenberg: Right.