BAREJA, Mr Mike, Director, Digital Technologies, AI, Cyber and Future Industries, Business Council of Australia [by video link] BLACK, Mr Bran, Chief Executive, Business Council of Australia [by video link] HARRIS, Mr William, Chief Operating Officer, Council of Small Business Organisations of Australia [by video link] CHAIR: We will resume our public hearing. Welcome. I understand that information on parliamentary privilege and the protection of witnesses has been provided to you. I now invite you to make a very brief opening statement, noting that anything further you can provide in writing directly to the secretariat; that way, we will get underway with questions from the committee members. Over to you for some opening remarks. Perhaps we'll start with the Business Council of Australia and then have COSBOA. Mr Black : We're very grateful to the committee for the opportunity to appear today. Personally, I apologise that I've been unable to be present in Melbourne. Thank you for accommodating me. We bring together the chief executives of more than 120 of Australia's largest businesses, and together they employ more than 1.1 million Australians. The first thing I'd say today is that AI is not some distant technology. In fact, most Australians are already using it every day, often without thinking about it. The navigation app that gets you around and the weather forecast you check each morning are good examples of that. Similarly, AI is already at work across our economy. CommBank is using AI to help cut fraud losses by more than 20 per cent. At BHP, it's being used to quickly spot oversized rocks and other things that might damage conveyors or hurt workers, meaning problems can be anticipated before they happen, improving safety and keeping iron ore loading. Medibank nurses are using the technology to deliver better wound care to patients in their own homes. Small businesses, as I'm sure we'll hear, are using it to do their books, write quotes and win back more time that can be dedicated to their customers. That all matters because Australia has a productivity problem, and, as we know, weak productivity ultimately makes the cost of living higher. AI can help people and businesses do more with what they have and deliver better services faster. We also know Australians have reasonable concerns about privacy, about scams and about what AI means for their job. Those concerns deserve straight answers. Jobs and Skills Australia's assessment is that AI is more likely to help people in their work than replace roles, and employers have a responsibility to bring their people with them, training them and involving them in how AI is introduced. That's why, for instance, the BCA's recent guide Adopting AI in the workplace was prepared. Ultimately, the question we consider for this committee is not whether AI comes to Australia in the distant future. Indeed, it's already here. The question is whether Australia is able to make the most of the extraordinary opportunity and manage the risks as they arise. Critically, we believe that means four things. Rules should address any genuine identified risks and not double up on laws we already have. We should prioritise skills development so all Australians can participate in the opportunity. We must prioritise one set of national rules for data centres, the power that AI needs. For our systems that matter most, being defence, national security and critical infrastructure, Australian control is key. If we can get those settings right, businesses will invest here, Australians will see the benefits in their daily lives, and more of the jobs, ideas and value will stay here in Australia. Mr Harris : I'll keep this brief. There are four main points I wanted to make today. The first is that, at a time, as Bran mentioned, that we find ourselves in the worst productivity growth decade in 60 years, AI provides us a potential get-out-of-jail-free card, if we get the settings right, and has the potential to significantly reduce the compliance and administrative burdens of running a small business. The second point is that, in the small-business community, AI adoption is broad but shallow, and what's really lacking is the practicality and depth of use. That, I think, is where we have the most potential to support small businesses. My third point is that I think we've done a disservice to small business in the way we've talked about AI in our media and political discourse and debate, swinging between either the robot apocalypse, 'Everyone's doomed,' or automating away all our problems, 'AI is going to fix everything.' What we've lost is the practical discussion of AI as a tool. I wouldn't use a hammer to put in a screw, and a lot of what's being sold currently as 'powered by AI' is using hammers to put in screws, so a lot of small businesses' first interaction with AI is negative. That's bred this culture of distrust in AI for small businesses. Again, it's that practical support through trusted sources that is lacking and needed. My fourth and final point is the importance of avoiding a kneejerk reaction to AI, particularly with how fast the technology is developing, and ensuring that we aren't overregulating and blocking access for small businesses to something that can really make a difference in their day-to-day lives. So, if we get the settings right for this going forward, I think we do have an actual once-in-a-generation opportunity to level the playing field for small businesses. I think that's why it's very important that we keep them in mind in any settings we're looking at going forward. CHAIR: Thank you very much to you both for those opening remarks. I want to touch on a couple of key points that you both raised and that we are hearing as a committee around this sense of trust and the ability to bring people with us. We speak about the potential for productivity uplift but, in my view, I don't see how we necessarily achieve that unless we can bring people with us—community members, small businesses and in particular workers. And this is my question to Mr Black. In your submission you talk about consulting workers about how it shouldn't operate as a general veto of technology or business decisions. As far as I'm aware, workers aren't asking for a veto; they're just asking to be told and heard and have a sense of agency as systems may seek to change how their rosters are set or how their performance is working. I'm just wondering why the BCA's starting point seems to be to rule it out for people who work alongside these systems. Can you explain that for the committee, please? Mr Black : I actually agree that the critical thing that we need to do in terms of the introduction of artificial intelligence into workplaces is to start with engagement and consultation with employees. That's precisely what we say in the guide that I referenced before that we prepared on integrating AI into the workplace. Indeed, what we've said in that guide and what I would really stress is that, as a business, you're always going to get the most out of AI if your proposal to integrate it into your operations starts with a conversation with your workforce as to what the problem is that you're trying to solve. Interestingly enough, some of the feedback I've had from companies, both here and overseas, is that they've found it in some instances difficult to realise productivity gains within AI at this early stage of its integration into their operations. Where they've done it best is where they start by having those engagements with their workforce to ask, 'What problems do you have where we could use AI to help you solve them?' So we would say: start from that point and then tailor your engagement based on the nature of the integration that you're proposing, and then make sure that along the way you're keeping humans not only in the loop but in control of key decisions that affect people day to day. That's what we've set out in the guide. I would say, in terms of your question, that we don't believe we need to see a change in the law at this stage. We think the law is appropriate and adapted to the types of circumstances we're seeing as part of the AI transition right now. But I would stress that from a business perspective you're always going to get the most out of the integration of AI into your operations if it's not just a case of bringing your workforce with you but they are there as part of the journey every step of the way. CHAIR: I appreciate that. We heard evidence earlier this week regarding where algorithms or AI are being used to purposefully look at efficiency gains and productivity and this gap that's starting to emerge between job as imagined and job in reality, taking into account the extent to which a maths problem or an algorithm can provide the quickest and fastest way to do a particular job, but then you put it in for a human to do that. The SDA put to us an example about Aldi logistics services, where it had required the workers to be told via their earpiece exactly where the fastest route was, but the timing of it didn't account for what was happening on the shop floor, whether something had happened and the thing wasn't necessarily in that place. The stress and added pressure that it was placing on those workers was pretty significant. On the face of it, it said that the efficiency gained from this new way of working was a huge uplift, but in reality it was just shifting the burden onto those workers in terms of psychosocial hazards and physical burnout. You speak of those guidelines and the importance of bringing workers with you. How do we engage in that way or be clear with businesses about the extent of just relying on the algorithm itself, and that gap between job as imagined versus job in reality? Mr Black : A big part of it comes down to talking about the case studies and examples of where AI has been introduced into workplaces well. One example that comes to mind, that stands out from within my own membership, is what CommBank have done in terms of frauds and scams; you may have heard a little bit about this when Matt Comyn appeared before the committee a couple of days ago. They're confronting real challenges day to day in terms of the number of fraudsters and scammers trying to take advantage of Australia's CBA customers. They've been able to, by directly engaging with their workforce on the problem of accounting for this increased challenge that we need to meet each and every day, increase their output, as I understand it, about twentyfold in terms of their capacity to detect and manage frauds and scams. At the same time, they've got the same number of people working within that division; they're just doing it far more efficiently and they are genuinely, as a consequence of that, being far more productive. They started with a problem and they've engineered a solution that's making their operations much better. There are so many examples I can reference that speak to precisely the type of operational improvements we've seen in that instance and we're also seeing right across our membership. I stress there that the more we can talk about those case studies and the more we can encourage people to think through that starting point—'What is the problem I'm trying to solve?'—the more likely it is that they're going to start with a conversation with their workforce. CHAIR: Thank you. Deputy Chair. Senator DUNIAM: Thank you very much for your attendance today. I think the Australian economy is at a bit of a crossroads with this emerging technology and the opportunities it presents. I wonder whether you might give us some guidance from a business sector point of view. What is required from government to provide the signal that there should be as much take-up as possible, and what support might be required from government in order to facilitate take-up of AI within commercial operations across the economy? What is required from a regulatory point of view? I don't know about the membership of the BCA, but I presume there are data centre operators and perhaps model developers who form part of your membership. What is required to provide certainty around growth of that sector? They're a couple of things I'm very interested in as we look to make recommendations to government about what might be needed to move from where we are now to where we ought to be. Mr Black : I would say a couple of things. You will no doubt have heard the expression 'the stack' used ad nauseum over the course of the committee's hearings. When I think about the spectrum of AI opportunities—and it truly is a spectrum—I think about it in the context of that stack. That stack, to my mind, is the development of chips, the development of large language models themselves, the infrastructure that is the data centres needed to undertake inference and model training. It's the skill sets that people have in order to best make the most of the AI opportunity. And there's the capacity to develop applications using artificial intelligence. Australia has got enormous potential across most of the stack. There are some areas where we're much stronger than others. I would make the following key points in terms of where we can enhance our strategy to benefit from this opportunity. First, with respect to chips, chips depend upon critical minerals, and we have those in abundance. Indeed, when you consider the critical minerals that are required to produce chips, I think we've got all of them save for boron. What that means is that we have enormous scope to benefit from the rollout of AI internationally, and it also gives us agency with respect to our engagement with other countries as well. What is important there is our scope to get these minerals out of the ground and to do it quickly. One of the key things that really stands out in that regard is the capacity for us to streamline our approvals processes at a federal level. We've talked about the criticality from an EPBC Act perspective of having these states accredited to administer the federal system. I would stress that that is going to be critical in terms of our capacity to benefit with respect to minerals. In terms of data centres, we would say that we've got to resolve that position in terms of copyright, and we would say we need to find a way through there that enables us to benefit from the investment opportunity that comes from being able to host centres that aren't just engaged in the inference—so responding to prompts and so forth—but are actively engaged in the training of models. That is really important. It's important in terms of investment, but it's also really important in the context of making sure that Australian data, particularly sensitive data, is retained on our shores, subject to our jurisdictions, subject to our laws. The next point I'd make is in terms of skills, I think we already have such an extraordinarily skilled population, and, critically, through our VET operators and through our universities, we have the capacity to adapt, reskill and upskill reasonably easily, or relatively easily, compared to many of our peers. The capacity for us to think through how we can deliver all people a base-level understanding and knowledge of how to engage with AI and how to benefit from it will be really important, particularly given that we still see that uptake rates are reasonably low across business. It's 12 per cent, and for large businesses it's only 35 per cent. There's a long way to go there. I think that skills component is really key. The last point that I would make is in terms of regulation—I think it's really critical that when we talk about possible regulatory intervention, we only do so in circumstances where there is an identified problem to solve—where we've genuinely undertaken a gap analysis, we've identified the gap and we look to solve it with a specific regulatory approach rather than broad sweeping powers. That's what businesses are looking for in terms of confidence moving forward. I apologise that was a long answer, but I hope that's been useful. Senator DUNIAM: Before I come to COSBOA, I just wanted to ask on copyright whether you've got any ready-made models of reform that we might be able to draw from to resolve that intractable problem? Mr Black : We've consistently argued for a copyright model, and I'll talk to that, but I might also just very briefly throw to my colleague Mr Bareja to provide some further reflections in that respect. But what we've said is that we want to, on the one hand, ensure that creatives have a scope to benefit from work that they create. And on the other hand, we need to make sure also that large language models have the capacity to be trained and trained on our content. We don't think it's practical for AI companies that are producing these large language models to have to engage with every content creator in order to secure a licence. We don't think that will work, and, ultimately, that will mean that Australians are worse off in the sense that there's less scope for models to be trained on Australian content. What we've proposed are arrangements whereby content creators should be able to opt out of having their content used as a basis for training large language models. We know the technology is available. We know that the technology continues to be developed and further worked up as well, which is really important. But we see that if you can have that type of opt-out arrangement in place, then, critically, that would mean that those creators that want to be recognised and compensated for their works have the capacity to do so in circumstances where licensing is entered into between them and AI companies on the one hand, and on the other hand, you would see the models themselves able to train on Australian content. We consider that to be very much in Australian's and Australia's interest. I might very briefly pass to my colleague for further elaboration. Senator DUNIAM: Sure. Mr Bareja : Thanks, Bran. You've pretty much covered everything there. I would say the only area of continued work that we would suggest is around how those opt-outs are conducted. This is an area which is ripe for more technological innovation. It's not a perfect system. No system is. But it's not something that should just be dismissed. It's something that should be worked on as a priority as to how we can get opt-outs working efficiently and transparently. Senator DUNIAM: Mr Harris, I asked a question earlier around what is required to facilitate take up across small to medium business enterprises and facilitate better and more productive use of AI. What's your view about what government could do to facilitate that? Mr Harris : This probably reflects a lot of what Bran was saying. I think the three biggest things are trust, accessibility and certainty. Trust is low, so I think government has a place in helping to build trust working through trusted sources like industry associations, advisers and accountants, particularly with a focus on practical support and getting people's foot in the door in the implementation of AI. Ideally, they'd go to each individual industry that is lagging in AI adoption and say, 'Here's a single practical use case for your industry that will save you X hours per week.' I think that would go a very long way. In terms of accessibility, accessibility of AI tools is quite high at the moment, but that's where regulation is a potential threat if we go down an overregulation path, particularly for small businesses. We want to maintain that accessibility. And then for certainty, as Bran said, there are things like existing regulations that largely covers any issues we're seeing arise from AI, but small businesses don't know how and where those regulations apply and what the situation is, even for simple things like if their AI customer service chatbot promises something that they don't want to deliver who's ultimately responsible for that. We might know the answer to that, but a small-business owner doesn't, and that, again, feeds into that lack of trust and hesitancy to engage with the technology. CHAIR: Ms Chaney. Ms CHANEY: Thank you for your evidence. Continuing on there with COSBOA, what are you hearing about AI driven scams at the moment in your membership? How would you measure or describe the level of readiness for that increasing risk? We are expecting to see an uplift of that with the current model versions. Mr Harris : I think this feeds into some of what I was talking about earlier about the way we're talking about the narrative around AI. The simple answer is that we're getting a lot of concern and confusion from small businesses in terms of AI enabled threats. The biggest factor there—and this is qualitative from our own discussions, to be fair—tends to be this uncertainty, again, around what those new threats are. They're hearing about the increase of all these new AI threats, but in reality it's largely the same threats we've always faced. There's just an increased exposure, whether it's putting more data into it or it's easier to develop a sophisticated scam or suspicious email. Once you get people across that, there's a lot less fear and you can engage with them, but I think it's because of the increased accessibility on the scammer's side, unfortunately, to develop scams and more suspicious emails et cetera. That's why we need a focus on entry level cybersecurity for small businesses as well, but we need to avoid becoming our own worst enemy and trying to make small businesses IT experts because that will scare them away from dealing with the issue. We need to say, 'Don't worry about these new fandangle threats. This is what a suspicious email is. Don't click the link. You don't need to know what a phishing email is. Suspicious? Done.' Ms CHANEY: So keeping it simple? Mr Harris : Yes. That can be surprisingly difficult. Ms CHANEY: Indeed. For the BCA, your submission includes a lot of case studies in which AI has been really effectively and successfully rolled out. Are there examples amongst your members—you don't have to name names—of AI being introduced that haven't been successful, and what can we learn from those experiences? Is that something that you have any visibility of? Mr Black : It's a really good question. The core bit of feedback that I've had from members goes to the point that I made earlier about being clear on the problems that you're trying to solve. What we have heard in Australia and also overseas, which has been really interesting, is that in the last two years particularly a lot of chief executives and boards have rushed to introduce AI just because everybody's doing it. They've given everybody an opportunity to have access to a Claude account or a ChatGPT account. They've expected to see productivity improvements as a consequence of that, but that hasn't arisen. What they have found, on the other hand, is that, where they're very clear on what problems are existing and what problems they therefore are trying to solve, they have been able to realise the gains. The caution that we've heard from chief executives and boards with respect to the introduction of AI is therefore very clear. Take the time to be clear on what problems you're seeking to solve before you integrate it into your workplace, and, as I mentioned before, once you start the process of integrating AI into your workplace, really engage with people who are using it on the ground day to day so you can get their feedback and make sure that the way you're thinking about your operations is entirely in line with what they're doing day to day. Ms CHANEY: What are you seeing in terms of take up of off-the-shelf proprietary models like Claude versus custom models that are built to solve a particular problem, as you just talked about? I'm really interested in understanding the extent to which access to proprietary models and those frontier models is going to be really important to Australian productivity gains or whether the real opportunities are in innovating Australian bespoke models that solve a particular problem. They don't necessarily have to be Australia, but those more bespoke models. Mr Black : It's a really good question, and, if you don't mind, I might ask my colleague to provide some reflections in that regard. He is engaged day to day in direct collaboration with our members on precisely these types of points. Mike, if you're able to comment that would be very useful. Mr Bareja : In short, all of the above. In Australia we need to be able to develop our own indigenous LLMs. We need to be able to have access to the best commercial models out there. We need to secure access to frontier models. These are all used for different purposes. In a couple of specific examples, I can think of a couple of those narrow train models that are already in use in Australia. You can imagine that across the university sector there are many universities and areas that are developing and training their own models for very specific research areas and doing amazing work. That should continue. Those don't require access to a frontier model, for example. However, on the other hand we have growing requirement to access to the best in the world for those national security, defence and critical infrastructure components and also for other specialised applications as well where the frontier is really necessary to make those, say, scientific gains where, otherwise, we are not going to be able to make as fast progress. We have an issues with how we can get or secure that access to frontier models. It's something that needs to be done with a strong eye for prioritisation. Where do those most critical and important applications lie? We should be using the frontier models for those particular applications. As I'm sure you've heard over the week, it's like using a Ferrari to drive down to the shops. You don't need that capability for most applications. Ms CHANEY: Yes. We've heard evidence that for business 90 to 95 per cent of business uses of AI will not actually need a frontier model and that the real benefit in those frontier models is likely to be in government functions like national security and cyber. Is that consistent with what you're hearing from members? Does that 90 to 95 per cent feel intuitively about right? Mr Bareja : That's absolutely true, but I think it's important to keep in mind that, among businesses large and small and the research sector, those frontier models are exceptionally useful for certain problems. I don't want to speak for my small and medium enterprise colleague, but imagine a small defence company that is working on designing a new aerospace component. The frontier model would be particularly important in being able to design aspects of that. But yes—overall, for most applications day to day, you don't need a frontier model. CHAIR: Ms Chaney, we might move on to the next set of questions. Ms CHANEY: I'll put questions on notice. Thanks. CHAIR: Brilliant. Senator Sheldon. Senator SHELDON: Mr Harris and Mr Black, there was a question raised about informing members of the association on what might be some good practices or better practices, and the submission goes through a number of proposed and suggested examples of better practice. What's your experience with better practice in small business? Have you got a code, and how widely is it being adopted or considered amongst your membership? Mr Harris : In small business, I'd say again that adoption is quite slow. But in terms of best practice, in my experience, the best approach is to do direct face-to-face—ideally—or online meetings with small-business owners and/or employees. The best approach is to work with their employees, to say, 'Okay, what are the painful aspects of your job or the bottlenecks day to day that you wish didn't take as long as they did?' and 90 per cent of the time that's just a conversation. Minutes is a really easy one; most people are familiar with that and say, 'I hate spending hours doing meeting minutes'. Okay; the question is then: is AI of help here? If the answer is 'yes', then the conversation might be, 'Let's explore how we implement it, and here's an example.' And then, once you get them into the ecosystem, they're more willing to explore and test other use cases. Getting them into the ecosystem is the really important part in the first instance. But it's essentially going with the approach of asking, 'How can we make your life easier on a day-to-day basis?' but also not trying to just say AI is a solution for everything. Because in my experience, when they raise those bottlenecks, I'd say it's probably something like 60 per cent of the time that, yes, there might be an AI implementation that can really help or save a bit of time. But 40 per cent of the time, no, it's not the right tool for the job, and it's, 'No issues—let's look at something else.' Senator SHELDON: I appreciate you giving some anecdotal examples. How have you gone with a rollout, or where are you up to with a rollout or roadmap for small business on adoption or utilisation of AI and associated technologies? Mr Harris : We have an informal proposal with the government on what could be done on a larger scale to do a rollout of AI. The simple answer, from our perspective, is that it's resource prohibitive as a not-for-profit to be able to do it on that scale. But what's really needed is that direct single application focus, we believe—of 'here's one use in your industry'. I think you could do it as simply as starting with the three or whatever industries we decide are lagging behind the most on AI, or what it's the most useful in, and give them that single concrete application. Then you build that out on an industry per industry basis. Senator SHELDON: That consultative process is showing some hope of being practical for the types of businesses you represent, which are quite broad in the small business arena. Mr Harris : Yes. Senator SHELDON: Mr Black, on that question of the actual rollout, and the acceptance and capacity for large and medium sized businesses as well, what are some of the practical pieces that the BCA can lend in conversations with the Small Business Association and the government on dealing with this question of social licence in the very broadest terms—business usage, government usage, community usage, workplace usage? Mr Black : It's a really good question. We are aware that there are concerns that people have, as I said in my opening comments. There are two critical points that I would make here. The first, as I mentioned earlier, is talking a little bit more about where AI is already just part of people's lives day to day. I think when people have that sense that it's still to come—that it's not here yet and that it's still to come—then that, in and of itself, causes concern. But, once you start talking about the different ways in which AI is already part of somebody's life, some extraordinary examples really bring this to life. We know that Medibank, for instance, is using AI to help manage wounds. Silverchain is using AI to enable doctors to really focus on their patients in consultations. Our large miners and our property companies are using AI to help manage safety on worksites, and we see companies like Telstra using AI to help predict how a bushfire might spread and therefore giving communities early warnings and the capacity to save critical infrastructure. That's, I think, a really good set of examples that describe very clearly and illustratively how AI is already here and part of people's lives. The more people accept that or understand it, I think, the more likely it is that they're able to accept what that means for them. I think the second component is really homing in on skills and giving people exposure to a base level of understanding and ability with respect to the use of AI. One thing that we're doing at the moment is working with government to try and create an online platform where businesses can put the internal programs that they've developed to try and upskill their own workforces. They can put them online onto this platform such that everybody has access to these types of upskilling and training programs. In turn, we hope that that will give people the chance to jump on and acquire at least a base level of understanding as to how AI can be used and how they specifically can use it in their lives from day to day. We think that this type of exposure will help bridge the gap, to the extent that it exists, in terms of building social licence. But I do think that skills are really critical in that respect. CHAIR: We're approaching time. Do you have a final question, Senator Sheldon? Senator SHELDON: Yes, on the question of the speed with which AI, as we all know, is here and is having substantial impact in the community—both opportunity and risk. Those risks to social licence definitely include concerns about job loss, and that's also affecting, I suspect, the uptake issue. When you look at the percentages in this country—we are 42nd out of 47 countries in believing that AI is a net detriment to communities—you see we've got a long way to go. You mentioned in your submission that there are existing laws that already mean that perceived problems, or problems that have occurred, can be dealt with. Are they sufficient, or is what we're doing sufficient, if we have to move so quickly? Are there ways of adding to those laws—I don't mean just in a regulatory sense, but I'm not excluding that—to actually encourage consultation, which allows better acceptance and opportunity for both business and the workforce—training, education and those sorts of things? Do we need to give an even deeper think to what we're doing? Mr Black : I do agree that we need to be thinking very deeply about this. I think that there are some processes that are already in place that are very useful in that respect, and I'll come to them. We don't think that at this stage there is a need for us to consider changes to the law in this regard, and we'd be concerned about going down this path in the absence of an empirical basis to show us that job losses are in fact a phenomenon and therefore a basis for real concern. What we are seeing here in Australia and overseas, across our membership and beyond, is that AI is not displacing roles; it is augmenting them. Of course, there are going to be situations in which there is a different outcome, but, in by far the majority of situations, that's precisely what we're seeing. That's what the evidence is showing. We know from our engagement with our members that they talk about AI changing the nature of roles people are performing but not replacing the roles. I think that's really important. As to your point in terms of deep thinking, we would agree with that. The simple reality is that there are unknowns in terms of where the technology will go and what that means right across our society, and I think what that points to is the importance of having processes in place between government, business, unions and other key stakeholders that can help us try and anticipate problems and then also respond and react and, if necessary, get ahead accordingly. This committee, I think, is a really useful example of that. But another really useful example is the work that Minister Rishworth is doing which proceeds on a quarterly basis at least for employers, unions and government to get together and understand the jobs market—where it's at right now, what the trends are looking like—and to have the opportunity to engage on what steps might need to be taken, moving forward. We entirely appreciate that, throughout this transition, for however long it will go—years or decades—we need to retain social licence along the way, and the way that we do that is by being honest and open about the problems, being honest and open about our level of understanding and genuinely engaging with workforces along the way. It's in our interests to be actively part of these conversations and ready to move if there is an evidence base that's presented that justifies that type of movement. CHAIR: Thank you very much. That brings an end to this session. Thank you to all of you for giving up your time today to participate and contribute to this really important discussion. Your experience is very valued. If there have been any requests for information or if you have any further information that you would like to provide to the committee, please do so through the secretariat in the next couple of weeks. Thanks again..