Latest / Herbert Smith Freehills Kramer Podcasts / On Just Terms S2 E11: Lawyers, risk, and the future of knowledge work with Professor Anthea Roberts, ANU Professor
Transcript
- 0:10SPEAKER_00: Today we're joined by Professor Anthea Roberts, professor at ANU and a leading scholar at the intersection of law, technology, and global risk.
- 0:19SPEAKER_00: We'll be exploring what it actually takes to work well with AI, not just adopting it, but developing skills, the mindset, and the agency to use it in a way that genuinely extends thinking.
- 0:31SPEAKER_00: Anthea brings a sharp and nuanced perspective on both the promise and the pitfalls, and we think you'll find this conversation as thought-provoking as we did.
- 0:40SPEAKER_01: Anthea, thank you so much for joining us today.
- 0:43SPEAKER_01: We wanted to start off with actually you just giving us a bit of an overview of your career today, because it's been such an interesting one.
- 0:49SPEAKER_01: You've been a litigator, you're an expert in international law, geopolitics.
- 0:53SPEAKER_01: Can you talk us through a bit of your background and then also how you then journeyed into your thinking and work around AI?
- 1:00SPEAKER_02: Sure.
- 1:00SPEAKER_02: So I think it's probably a relatively unusual background.
- 1:04SPEAKER_02: I started in law, my only law job in Australia was to be an associate at the High Court for the Chief Justice.
- 1:10SPEAKER_02: From there I really branched into international law because I was really interested in how law looked different from different countries.
- 1:16SPEAKER_02: So I did that both through study but also through working in a law firm in New York and London.
- 1:21SPEAKER_02: But I'd always had a passion for academic research and I always thought I would do a combination.
- 1:25SPEAKER_02: So I jumped across into the academy, first teaching international law and trade at LSE, then Harvard and Columbia.
- 1:30SPEAKER_02: But all of that got me to a point where I started to think many of the issues that I was facing really required going beyond the law.
- 1:37SPEAKER_02: So I started thinking in much more interdisciplinary ways about economic globalization or the pushback against globalization, which brought me back to an interdisciplinary professorship here at ANU.
- 1:46SPEAKER_02: And yet that took me through to thinking about complex policy making and decision making, which in turn took me to AI.
- 1:52SPEAKER_02: And so I think it would be fair to say that it's not a traditional escalator career.
- 1:56SPEAKER_01: We've been reading a lot about some of your work, particularly on the future of AI and ways to be thinking about AI.
- 2:03SPEAKER_01: One thing that I picked up on, which I think might resonate with some of our listeners, is I think you've described this moment in time that we're at now as kind of one of the scariest, but also one of the most exciting.
- 2:14SPEAKER_01: So can you talk us through how you're seeing it from both of those sides?
- 2:18SPEAKER_02: So it's really interesting.
- 2:20SPEAKER_02: You're really clearly seeing a different psychological reaction from people to AI because it is so disruptive, I think.
- 2:26SPEAKER_02: And some people tend to approach it very much with a threat lens of this undermines my identity, this undermines all of the skills that I spent all this time developing, this undermines my economic worth, and what does it mean to my job?
- 2:38SPEAKER_02: And so you very clearly see this threat perception.
- 2:41SPEAKER_02: And on the flip side, you also see some people have a real opportunity of like, wow, what can I do now that I couldn't do before?
- 2:48SPEAKER_02: And that's not just a question of how could I do the same things faster, but how could I do something that I was unable to achieve previously, or how can I branch out to scaffold myself with new skills?
- 2:59SPEAKER_02: And this kind of difference in personality type as to whether you approach this disruption through a threat lens or through an opportunity lens, I think is going to have a massive impact on the way that you posture and the way that you adapt to this particular change that's that's affecting all of us.
- 3:16SPEAKER_01: And picking up on that, what's one of the things, I suppose you're doing a lot of work at the moment, what's one of the most exciting things that you're looking at or interested in?
- 3:25SPEAKER_02: So I think we've always been interested in not how do you use AI to automate low-end tasks, but we're incredibly interested in how do you use AI to augment humans to do the kind of thinking that they weren't able to do previously.
- 3:40SPEAKER_02: And so it in that sort of trajectory I explained about my own career, I'd really got to the point of thinking about these complex policy challenges where you had people from backgrounds in economics and in security and in environment who were all trained in their own silos and specialties because as humans, we can't cover the field.
- 3:59SPEAKER_02: And when we were dealing with wicked problems, they were sort of coming together and colliding.
- 4:03SPEAKER_02: And we really didn't have good structures for interdisciplinary analysis.
- 4:06SPEAKER_02: If you look at universities, we tend to be in our silos.
- 4:09SPEAKER_02: If you look at government departments, we tend to be in our silos.
- 4:12SPEAKER_02: And so I was really interested in how do you actually create structures and ways of thinking that allow you to synthesize and work through complex problems across these many different levels.
- 4:21SPEAKER_02: This is just something that humans completely are struggling to do.
- 4:25SPEAKER_02: But it's something where with the use of the right tools and the right AI techniques, we can actually get AI to help us do a kind of thing that we can't do by ourselves.
- 4:34SPEAKER_01: It's interesting when you talk about these areas.
- 4:37SPEAKER_01: When I think about lawyers, we're not always known as one of the most naturally innovative types.
- 4:42SPEAKER_01: You know, we're analytical people recognize that.
- 4:44SPEAKER_01: We're very thorough and detailed focus, but we're not necessarily always perceived to be at the forefront of new initiatives and innovation.
- 4:54SPEAKER_01: But was which was then when I was looking at some of your reading, it was interesting to me that you see that kind of the law is the gateway through to this new phase of AI-driven transformation.
- 5:05SPEAKER_01: So I wanted you to kind of unpack that for us and explain why you see kind of law as this kind of gateway.
- 5:11SPEAKER_02: Yeah, it's really counterintuitive, isn't it?
- 5:14SPEAKER_02: It's not what you would have expected.
- 5:15SPEAKER_02: So what we're seeing is that it's affecting different fields at different points in time.
- 5:20SPEAKER_02: And so ground zero has really been software.
- 5:22SPEAKER_02: And we're really seeing that through the use of code and coding agents, and coding agents are the first all-purpose agents we've seen that we've really seen that have a dramatic effect in software.
- 5:34SPEAKER_02: But the next one after that is going to be knowledge work.
- 5:36SPEAKER_02: And so the question is, where does it come first in knowledge work?
- 5:39SPEAKER_02: And if you look at where there have been the huge investments, some of the biggest ones have been into law, some of the biggest valuations have been into law with things like Harvey and Lagora.
- 5:49SPEAKER_02: And there are a number of reasons that make law really well suited for this.
- 5:52SPEAKER_02: So one is that it's very textual.
- 5:54SPEAKER_02: So it is literally the code, but code in English language.
- 5:58SPEAKER_02: And so the systems are very well suited to it, that you have repeated formats in many ways, so NDAs or contracts.
- 6:05SPEAKER_02: The individual words matter, but there are a lot of sort of repeat patterns.
- 6:08SPEAKER_02: But interestingly, I also think it's because of this structural position.
- 6:11SPEAKER_02: So if you look at most lawyers, some of them are doing just core law, but many of them are actually functioning with much broader risk and strategy profiles, either in their firm advising corporations or as general counsel within corporations.
- 6:27SPEAKER_02: And that means that the lawyers often sit not just in law, but they in risk and strategy and have to balance these different things, and that they are connected to every part of their firm internally, but also they are the ones that connect to other firms externally through deals and through contract negotiations.
- 6:42SPEAKER_02: And so what this has actually meant is that the huge amount of money that has gone into legal AI actually is treating legal AI, I think, as a beachhead into professional services AI.
- 6:53SPEAKER_02: And so you're seeing ones like Harvey that are not just working out how do we bring it into law firms, they're working out how can you get those law firms to then share it with their clients.
- 7:02SPEAKER_02: And then how can you get law firms to work with other non-legal firms like PE firms in various deals?
- 7:09SPEAKER_02: And so they're setting up the infrastructure for all of that.
- 7:11SPEAKER_02: So you can actually see law in that case becoming like the central node of infection within a company and between companies.
- 7:18SPEAKER_02: And this is totally counterintuitive when you think legal tech has always been a small and sleepy and backward area, and yet now it is getting outsized investment.
- 7:28SPEAKER_02: So if you if you look at valuations of Harvey, for example, at 11 billion, that's 10 times ahead of where the accounting firms are at.
- 7:35SPEAKER_02: And that's a really surprising shift, I think, in this field.
- 7:38SPEAKER_00: Something I wanted to expand upon from a comment you made earlier, which was really interesting about how AI can fuel a better level or a heightened level of critical thinking or at least help us to think more critically, particularly across disciplines.
- 7:52SPEAKER_00: My observation of the legal profession, perhaps in part corporate Australia, is reactions to AI have tended to fall into two camps.
- 8:00SPEAKER_00: One is this is going to save a lot of time and money, and that's great.
- 8:03SPEAKER_00: The time and money that we will spend on discovering 200,000 documents in a piece of litigation is going to go down.
- 8:10SPEAKER_00: But there's another, perhaps more engaging bucket, which is how can this help us as lawyers think proactively, collaterally, approach problems in a unique and different way?
- 8:21SPEAKER_00: I also see that you've said something similar about in a more general sense about how it's helped you think harder, think more critically.
- 8:28SPEAKER_00: I just wanted to engage with you on, based on your experience, how that's happened, because I think there's a bit of a misconception that AI is just about doing a timesheet more quickly.
- 8:37SPEAKER_00: And it's a much more fundamental change than that.
- 8:39SPEAKER_02: Yeah.
- 8:39SPEAKER_02: And like I think at the moment it hasn't been a more fundamental change for most workplaces.
- 8:44SPEAKER_02: They've had so many governance layers and ICT layers that most people have been, A, given access to really poor quality tools, I would say.
- 8:52SPEAKER_02: So Copilot is the one that's used the most, but it's sort of two years behind the frontier.
- 8:57SPEAKER_02: And also it's been kind of added in as like small additional things on your existing processes.
- 9:02SPEAKER_02: Whereas if you actually really want to embody this, you actually need to completely redesign your processes with AI first.
- 9:09SPEAKER_02: And you need to develop the techniques of people about how do you use this to do that higher-end thinking work.
- 9:14SPEAKER_02: So just to give you an example of what we do, uh we we spun this out into a company called Dragonfly Thinking.
- 9:21SPEAKER_02: Dragonflies have the many different lenses that they incorporate into a single vision, and that's where we got the concept from.
- 9:26SPEAKER_02: So in any of our strategic intelligence work, we have Dragonfly co-create with us as a lead orchestrator.
- 9:34SPEAKER_02: And Dragonfly itself then has around 15 different specialist agents, and each of those agents has about a dozen specific skills, and we co-create with the orchestrator to work out which agents to call on, which skills, and it's a collaborative endeavor.
- 9:50SPEAKER_02: And then we can turn it into completely different workflows.
- 9:52SPEAKER_02: So one of the workflows I've got running in the background at the moment has 22 different agents, and it takes somewhere between five and 24 hours to run this particular process.
- 10:01SPEAKER_02: And the sheer intellectual weight of that is really beyond anything I've experienced before to be able to run so many different agents in this sort of time period.
- 10:12SPEAKER_02: And it's not something we've really seen hit knowledge work, I think, in Australia yet.
- 10:16SPEAKER_02: So most people, I think, haven't crossed that particular chasm.
- 10:20SPEAKER_02: I think they're still very much thinking about how do we automate existing tasks, not how do we fundamentally rethink.
- 10:26SPEAKER_02: And in order to fundamentally rethink, you don't just need the AI skills and the AI agents.
- 10:30SPEAKER_02: You actually need to get your substrate right.
- 10:32SPEAKER_02: So a lot of where AI doesn't do well at the moment is where it doesn't have sufficient context.
- 10:37SPEAKER_02: And so you need to get all of your context together in one place, organized and machine readable.
- 10:42SPEAKER_02: So what we've done, for example, in the firm is we're capturing that we're doing, getting it all in Markdown and JSON files, all machine readable.
- 10:51SPEAKER_02: And then we actually have AI agents work alongside us.
- 10:53SPEAKER_02: So literally in our Slack channels, we have us communicating with each other, but we have the agents who are also communicating with us and in some cases communicating with each other.
- 11:02SPEAKER_02: And it's it's a very new way of starting to think of how do you go AI native?
- 11:07SPEAKER_00: Wow.
- 11:07SPEAKER_00: So that's layer upon layer of AI working sort of collaboratively with humans to develop a new way of thinking that's essentially, my words, cross-platform.
- 11:16SPEAKER_00: And so I'm interested in like, could you develop that a little bit maybe from Dragonfly's perspective and just thinking about what the future holds for us as lawyers in this space?
- 11:26SPEAKER_00: How can we think of AI as a thought partner, something not to be fearful of, but to embrace and perhaps to elevate our own legal service providing skills?
- 11:35SPEAKER_02: Yeah.
- 11:35SPEAKER_02: So I'll give you an example of one that we've actually done with an Australian firm.
- 11:39SPEAKER_02: They were looking at the Australia-China relationship and what it might mean for regulation going ahead in areas like foreign direct investment.
- 11:46SPEAKER_02: And that's really got a very core legal element to it, but there's also a lot of political dynamics and sociodynamics, and you're really dealing with a very complex interconnected system.
- 11:57SPEAKER_02: So they had actually internally developed a dossier of over a hundred events in the Australia-China relationship, which they'd put forward in a categorized way and in a linear way.
- 12:06SPEAKER_02: We working with them were able to take that and then turn it into a complex systems map of who the main actors were, how they were viewing the problem, how they were connected to each other in positive and negative ways with similar views or different views or tensions.
- 12:20SPEAKER_02: We were then able to identify all the main drivers that seemed to be behind the dynamics and how they were interacting with each other's to create like reinforcing dynamics or balancing dynamics.
- 12:30SPEAKER_02: From there, we were able to take that into different scenario projections of how might this unfold into more positive and more negative futures.
- 12:37SPEAKER_02: And from there we could then assess with different interventions what impact that might have on the system or how you might respond.
- 12:43SPEAKER_02: So this is a much broader spectrum of things that are totally relevant if you are a law firm or a general counsel and trying to advise your company on risk and strategy.
- 12:55SPEAKER_02: And it goes well beyond the law, but also well beyond what a single human can hold in their head.
- 12:59SPEAKER_02: And that's the sort of thing that we do at Dragonfly.
- 13:01SPEAKER_02: And then of course we kept human in the loop at the beginning and at the end of that process to check that we were confident in the answers and what was put out.
- 13:08SPEAKER_02: I think we've actually got a version of that advice actually up on our website if you want to see it.
- 13:12SPEAKER_02: But it just gives you some sense of how you can push into these new domains.
- 13:19SPEAKER_00: Yes, it's a legal lens initially, but really it's a diagnostic tool almost to predict the future or risks that might emerge in the future.
- 13:26SPEAKER_02: Right, absolutely.
- 13:27SPEAKER_02: And in a place like Australia, but also in the US, where I also teach, a huge number of people who do law degrees don't end up as traditional lawyers.
- 13:36SPEAKER_02: So it's a very good critical thinking set of skills.
- 13:38SPEAKER_02: And I think going forward, that ability to have that deep domain expertise, but also that breadth to be able to go into different areas, AI can really supercharge that.
- 13:48SPEAKER_02: So it can scaffold you to go into new areas, in my case, software engineering and AI expertise.
- 13:53SPEAKER_02: But it's also one that's deeply furthered by having that deep expertise in the T because you're able to get better outputs out of the AI than the AI by itself or a generalist using the AI by itself.
- 14:06SPEAKER_02: But that's only if you have the deep domain expertise and you multiply it by your AI skills.
- 14:11SPEAKER_02: And what we're finding at the moment is often the more senior people have the deep domain expertise, but not the AI skills.
- 14:16SPEAKER_02: Or we're seeing more junior people have the AI skills, but not the deep domain expertise.
- 14:20SPEAKER_02: But the winners going forward are going to be the ones that develop really strong ones in both and that they compound together.
- 14:26SPEAKER_00: Understand.
- 14:27SPEAKER_00: And part of that sounds almost generational, that if we're having this discussion in in a generation's time, I know things will change radically, but there'll there won't be this concern that there's a lack of AI skills.
- 14:38SPEAKER_00: Everyone will have that.
- 14:39SPEAKER_00: And so eventually these should just become embedded in our way of thinking.
- 14:42SPEAKER_02: I think that's right.
- 14:43SPEAKER_02: But I I think part of it is generational.
- 14:45SPEAKER_02: I think part of it is segmented by time.
- 14:48SPEAKER_02: So I think young people often have time to learn AI skills, but don't yet have the domain expertise and deeply worry that they might not develop them.
- 14:55SPEAKER_02: I'm actually finding that a lot of very good people who have recently retired actually also have the time to learn AI and they're doing really wonderful things.
- 15:04SPEAKER_02: It's the people in their 40s and 50s that are at the top of their sort of profession, but very, very time poor, that literally don't have the cognitive bandwidth to learn the new skill.
- 15:15SPEAKER_02: I think they're actually in some trouble.
- 15:17SPEAKER_02: But if they take the time to learn it, what they're able to get out of it is so much more, I think, than a new grad that just has the AI skills.
- 15:25SPEAKER_02: And so I would really encourage all of your listeners to really think like this is the most significant skill any knowledge worker is going to have to learn in their entire career.
- 15:35SPEAKER_02: And I think it's better to get on that journey earlier rather than later because it is a skill, it is a technique, it takes time to develop.
- 15:42SPEAKER_02: And I've seen that in my own company and that different people not just have different time to be able to do it, but different psychological predispositions.
- 15:49SPEAKER_02: So if you're somebody that's got very high agency and very high curiosity and doesn't mind experimentation and failure, you will actually do a lot better with this than people that are more traditional, sort of, I want to know exactly what I'm doing and what the output's gonna be and what the return on investment is going to be.
- 16:05SPEAKER_02: And so there are differences in psychology that are playing out in this case as well.
- 16:09SPEAKER_00: I'm definitely going to blame my lack of time for my lack of cognitive bandwidth going forward.
- 16:14SPEAKER_00: You've given me a good idea.
- 16:15SPEAKER_00: It's really true.
- 16:16SPEAKER_00: I've been accused of that before for different reasons.
- 16:18SPEAKER_00: Just before I hand back to Mel, one more general question.
- 16:21SPEAKER_00: In the area that you're developing at the moment through Dragonfly, and perhaps in the area of the legal profession more generally, are there a few or common misconceptions about AI that we should just debunk while we're together, just so that our audience can know that there's a couple of things that they're that they've been thinking about incorrectly, perhaps?
- 16:40SPEAKER_02: So I think the thing that stands out to me the most is how many times people that come from word backgrounds, and so lawyers tend to be like word background people, think that they can't do this because they think that it is coding, they think it is technology, they think it is like ones and zeros.
- 16:56SPEAKER_02: And actually, the people that are best able to develop this expertise are often subject matter experts that are very good with language and are very, very clear communicators and that can develop a really clear theory of mind for how the AI is reacting and why it's doing what it's doing so that they can direct.
- 17:13SPEAKER_02: So if you're a very, very good manager of people, you can often become a very, very good manager of AI agents, even if you have no technical background.
- 17:21SPEAKER_02: And so I think at this stage, a lot of like lawyer people think it's not for me because I'm not technical, but actually you probably have the skills that would make you better than a software engineer at it, but you have to lean in and try.
- 17:35SPEAKER_02: And here, your ability to communicate in language is so important.
- 17:40SPEAKER_02: And lawyers are very good at communicating with language.
- 17:44SPEAKER_02: And so I think there's a real advantage here that many people are not grabbing, that they should be grabbing.
- 17:48SPEAKER_00: Yeah, it's a great observation.
- 17:50SPEAKER_00: I don't think it would have been that obvious to many lawyers.
- 17:53SPEAKER_00: So that that's very insightful.
- 17:54SPEAKER_00: Mel, over to you.
- 17:56SPEAKER_01: Well, but before I finish up on another topic, I wanted to touch on, it just shows that really the importance of having time, having commitment from whether it be firms or the clients that we're working with, giving people time, committing to the investment is so critical to this.
- 18:11SPEAKER_01: But maintain the flexibility in thinking about the way that we're doing things that, you know, as you as you pointed out, Anthea, that the grads have been doing for a shorter period of time and so maybe are more naturally flexible, given they're not so set in their ways.
- 18:24SPEAKER_01: But the importance for more experienced practitioners who, as you say, have the legal knowledge and the experience, but who may be more set in the ways that they like to do things, the real importance of maintaining that flexibility in thinking and in approach, because I think it also helps you learn more quickly.
- 18:41SPEAKER_01: And just when you're touching on language there, the one thing I've been playing around a little bit at the moment with Lagora is this ability to work with Lagora to help improve your prompts.
- 18:50SPEAKER_01: And so just the learning process and the back and forth between the tool and myself, as I think is used, just you have described as a thought partner, you know, to try and and through that process, myself learning, okay, this is how I should be framing it here.
- 19:04SPEAKER_01: This is how it wants me to explain things to it, okay.
- 19:08SPEAKER_01: It's really fascinating.
- 19:10SPEAKER_02: And so I'd say a couple of things to that.
- 19:11SPEAKER_02: So that's that's exactly one of the things that people need to learn to develop.
- 19:15SPEAKER_02: So one of the top predictors of how you go with your AI skills is that iterative quality of going back and forth.
- 19:20SPEAKER_02: So not just thinking it's an answer, but going back and forth.
- 19:22SPEAKER_02: But do you see what you're actually developing there is you're actually developing not just the iterative dynamic, but also a theory of mind for how do I need to communicate so that it understands what I want.
- 19:32SPEAKER_02: And what the other thing you're doing is you're you're really exercising something that we call learning agency, which is often we've had an expectation with computers because we're not technical, that when something goes wrong, it's like computer says no and you're kind of helpless and you can't do anything about it.
- 19:46SPEAKER_02: But actually, there's a huge amount you can do about it with this type of computer because you can ask the computer.
- 19:51SPEAKER_02: And so one of the ways I think about it is that there has never been a technology in history that it makes it easier to cognitively outsource and cheat on your homework.
- 20:00SPEAKER_02: And there has never been a tool in history that makes it easier to cognitively extend and learn new things.
- 20:06SPEAKER_02: But you have to be the one that shows the agency to actually do the kind of extension and learning that you're going through at the moment.
- 20:12SPEAKER_02: And once that unlock happens, I'll bet you you're not going to just do it for prompts.
- 20:16SPEAKER_02: You're going to use it to teach you all sorts of things.
- 20:19SPEAKER_02: And then to go to Jason's point, you'll end up finding young people will get you to do it for completely different things.
- 20:25SPEAKER_02: So I'm obviously an advocate of using AI for many things.
- 20:28SPEAKER_02: And I try to encourage my kids to think about it in really critical ways.
- 20:32SPEAKER_02: But just on this weekend, my daughter was like, well, if you want me to use AI, take a video of this dough that I've been making and get it to analyze like why it's not working.
- 20:39SPEAKER_02: And I was like, really?
- 20:40SPEAKER_02: And she was like, You just try it.
- 20:42SPEAKER_02: I'd never done anything like that.
- 20:43SPEAKER_02: Sure enough, we took the video, we took it in.
- 20:45SPEAKER_02: The AI couldn't watch the video by itself, but it could slice it into frames, and then it gave an analysis of what was happening with the yeast and to try this and to try that.
- 20:53SPEAKER_02: And we went back and forth.
- 20:54SPEAKER_02: And so again, it's just treating it with learning agency, expanded it for my daughter.
- 20:59SPEAKER_02: It also expanded it for me.
- 21:01SPEAKER_00: How did the finished product come out?
- 21:02SPEAKER_00: I mean, that they leave us hanging.
- 21:04SPEAKER_02: So we went through two rounds of revisions and it didn't end up working.
- 21:08SPEAKER_02: And then we jointly decided with the AI that we would scrap that one, and then we made another one again.
- 21:13SPEAKER_02: And it was awesome next time around.
- 21:15SPEAKER_01: There you go.
- 21:16SPEAKER_01: Yes, lot the iterative, iterative process.
- 21:19SPEAKER_01: I like it.
- 21:20SPEAKER_01: So obviously you're doing a lot of work on AI.
- 21:22SPEAKER_01: There's a lot of resources that are available online that we might even link for our listeners who want us to learn more.
- 21:28SPEAKER_01: But as I was digging through a lot of those, Anthea, one thing I saw was some work that you've been doing or thinking around risk management, which is obviously something we do.
- 21:37SPEAKER_01: It links to your interdisciplinary and dragonfly thinking that you were alluding to.
- 21:42SPEAKER_01: But is this concept of obviously law firms, organizations, our clients want to manage and mitigate risk, but sometimes there being an unintended consequence of that that you end up minimizing reward or opportunity.
- 21:56SPEAKER_01: And so you've done some thinking around that of how you can actually try and shift.
- 22:00SPEAKER_01: The balance so that you're getting the right outcome.
- 22:02SPEAKER_01: And I thought I'd be interested to hear a bit more about on this topic from you.
- 22:06SPEAKER_02: So interestingly, this was the one that first took me into the AI.
- 22:10SPEAKER_02: So I had been doing work with some of the different government departments when they were thinking about Australia-China relations and how to develop regulations or develop laws.
- 22:21SPEAKER_02: And there you saw a very strongly security-minded group that very much focused on the risk of engagement with China.
- 22:27SPEAKER_02: And you saw a very strongly opportunity reward-focused group, The Economists, who very much saw the opportunities of this engagement.
- 22:35SPEAKER_02: And they were in their own communities and they didn't really talk to each other.
- 22:38SPEAKER_02: Or when they did talk to each other, they generally didn't like each other very much.
- 22:41SPEAKER_02: They had very different frameworks and they often thought each other were either stupid or malevolent.
- 22:45SPEAKER_02: And so I started working on how can we hold together risk and reward?
- 22:49SPEAKER_02: And what are the elements of those and how do they relate to each other for exactly the reason you say, Melissa, which is often in your attempt to minimize risk, you're also minimizing reward.
- 23:00SPEAKER_02: And if you want to be strategic, you actually need to balance the two.
- 23:03SPEAKER_02: And AI adoption is a classic case of this.
- 23:05SPEAKER_02: If you only focus on minimizing the risk, you actually won't get a computer, you won't turn it on, you certainly won't connect it to the internet, and you would never give access to an agent on it.
- 23:14SPEAKER_02: Whereas many of the rewards come from doing all of those things.
- 23:17SPEAKER_02: And so actually trying to hold this balance between risk and reward is really useful.
- 23:21SPEAKER_02: What I found in trying to do that was that it was actually really helpful to stop just thinking about risk and reward about an individual decision at an individual point in time, but to actually introduce this third element of resilience, which is how do you think systemically and over time?
- 23:34SPEAKER_02: And under resilience, I really had three different categories of when something happens to a system, you can either absorb the shock, so you can take a punch and not be injured, or you can have like a run-on PPE or vaccines, but you've got to stockpile, so it's okay.
- 23:48SPEAKER_02: So you can absorb and remain on change.
- 23:50SPEAKER_02: You can adapt, so you can do small-scale minor modifications to adapt and keep going.
- 23:55SPEAKER_02: So for example, in Australia, when you were doing the lack of PPE, you could then start to try and source it locally or build up various alternatives.
- 24:04SPEAKER_02: And then the final one is actually transformation, which is not minor incremental change, but how do you actually fundamentally pivot to redefining your systems, redefining your incentives, redefining your processes so that you might get to a more sustainable place going forward?
- 24:18SPEAKER_02: And actually, if you minimize risks too much, you actually often undermine your resilience to be systemic and over time.
- 24:25SPEAKER_02: And if you maximize your rewards too much, you actually do the same.
- 24:28SPEAKER_02: So often if you're maximizing your short-term rewards, you're often doing it at the detriment of the kind of investment you need to make into research and development or into training or into thinking of alternative business models that you might need for adaptive or transformative capacity.
- 24:40SPEAKER_02: And so I ended up developing it as a bit of a system to always be thinking about not just risk and reward, but what impact are they going to have on the resilience of my system over time.
- 24:48SPEAKER_02: And that could be the resilience of me personally, of my business, of my country at different, at different scales.
- 24:54SPEAKER_02: And it's a it's a beautiful case of interdisciplinary thinking, but also one where it was really helpful to have AI.
- 25:02SPEAKER_02: And so we did the first pilots into the federal government, and we did it based on the risk reward and resilience framework because it was relevant to many of the issues that they're facing on everything from critical minerals to EVs.
- 25:13SPEAKER_02: And I think it's relevant to many of our firms as well.
- 25:17SPEAKER_01: A bigger topic than we can hope to grapple with or educate everyone on today.
- 25:21SPEAKER_01: But do you have a couple of practical takeaways or tips or tricks for us, but also for our clients, we're trying to find this balance through the risk reward framework that you've described.
- 25:33SPEAKER_02: Yeah, so I would definitely say that if you narrowly think about minimizing risks, you will not adapt properly to the AI technology because you will be too restrictive.
- 25:43SPEAKER_02: But also if you narrowly focus on return on investment and immediate rewards, you're also not going to adapt properly.
- 25:51SPEAKER_02: You actually need to give people real time for adaptive play to work out what's happening.
- 25:56SPEAKER_02: And you need to give some people a chance to really think about how we would transform our services to put AI at the center of our operating system rather than just tacking it on to what we're already doing.
- 26:08SPEAKER_02: And so I think that many people are thinking about minimizing risk or maximizing these short-term rewards, but we need to be thinking actually over time, what's our absorbative capacity, what's our adaptive capacity, and in particular, what's our transformative capacity.
- 26:22SPEAKER_02: And I don't think that we're thinking in those terms enough yet as organizations.
- 26:26SPEAKER_02: And I think it's going to bite us because what I see is an evolving kind of K-shape where some people who are open to and exploring with AI are starting to see compounding benefits in terms of what they're able to do.
- 26:39SPEAKER_02: And other people are going to be falling behind.
- 26:41SPEAKER_02: So you're going to have a K-shape, I think, at the individual level.
- 26:44SPEAKER_02: I think you're also seeing it at an organizational level because I actually think many big companies really struggle to onboard AI.
- 26:51SPEAKER_02: So smaller companies and they're much find it much easier to go AI native.
- 26:55SPEAKER_02: And I also think you see it at a country level.
- 26:57SPEAKER_02: So Australia, for example, has good consumer adoption of AI and a lot of AI governance, but it's actually got very low organizational adoption of AI and very low return on investment.
- 27:09SPEAKER_02: And so on all the things where we could actually really be getting the rewards from it, we're not, because we're very focused on things that are free or things that are low in risk.
- 27:18SPEAKER_02: And this is such a disruptive time when I think we need to be thinking more broadly than that.
- 27:22SPEAKER_01: Been such a fascinating discussion.
- 27:24SPEAKER_01: I think we'll we'll end it there for today.
- 27:26SPEAKER_01: But thank you again, Anthea, for taking the time to speak with us.
- 27:29SPEAKER_00: We appreciate it.
- 27:30SPEAKER_01: Thank you for the insights.
- 27:31SPEAKER_01: Lovely to see you guys.
- 27:32SPEAKER_01: Thank you.
- 27:33SPEAKER_01: What stands out from today's conversation with Professor Anthea Roberts is three key things that I think are important for us to prepare for the next phase of AI in our profession and for corporate Australia.
- 27:45SPEAKER_01: First, the importance of not viewing AI through a threat lens and trying to reframe our thinking through an opportunity lens.
- 27:52SPEAKER_01: Many of us are already there, but I think there's still some work to do.
- 27:55SPEAKER_01: Second, investing the time it takes to develop the skills required to get the most out of whatever AI tool or agent that you're using.
- 28:04SPEAKER_01: And third, being really flexible in the way that we approach this new horizon and staying open to new ideas, new ways of doing old tasks.
- 28:13SPEAKER_01: Anthea talks about a fundamental rethink in how we approach our work.
- 28:18SPEAKER_01: And I think this really means it's the difference between using AI to cognitively outsource work versus those who are cognitively extending.
- 28:27SPEAKER_01: And what's clear from the discussion with Anthea is that it's those who are in the ladder cab that will really accelerate in business and in the profession, and those who will be really the best type of trusted advisor for their clients.
- 28:40SPEAKER_01: Thank you for joining us for this episode of On Just Terms, and we look forward to continuing the conversation.