Good afternoon and welcome to our lunchtime conversations from the International Labour Conference here at the Palais des Nations in Geneva. These conversations are brought to you by the ILO's Future of Work podcast series. I am Ibon Villelabeitia. Over the coming days, we will be exploring key 4 00:00:22,960 --> 00:00:30,400 issues shaping the world of work with experts from across the globe. Today we focus on one
of the defining forces of that change. Artificial intelligence is already reshaping work from jobs to skills to everyday working life. But the key question is how this will unfold. Will it strengthen decent work, dignity, and share prosperity or deepen inequality and exclusion? The Director-General of the ILO describes this as a moment of choice. The outcomes are not predetermined but depend on governance, institutions and social dialogue. In this
session, we will explore what this means for workers, employers, and economies, and what a human-centered future of work requires. To examine these questions, we are joined by four distinguished experts from the ILO and academia. We have Kostas Papadakis who is a special advisor, the Labour Governance and Sectoral Policies department at the ILO. Thank you, Kostas. We also have Janine Berg, a senior economist in the Research department of the ILO. We also
have Hannah Leipmann who's also an economist in the Research department of the ILO. And finally, we have Valerio De Stefano who's Professor of Law and Canada Research, Chair of Innovation, Law and Society in Osgoode Law School in York University in the city of Toronto. Thank you all for being here with us. Kostas, let me start with you for the big picture. As I said, the Director-General describes this as a moment of choice for AI and work. What is the key message behind this framing and
why is this moment so important for the world of work right now? Yes, thank you Ibon and thank you for the introduction. Indeed, as you said in your introduction, the key message of the ILO Director-General to the International Labour Conference through his report for the first time dedicated to the question of artificial intelligence is that outcomes, positive or negative, associated with this new technology are not predetermined. Whether artificial intelligence will lead to
increased productivity and to helping workers or on the contrary to increasing inequalities and insecurities will depend on the choices we are making and these choices are of course going through policies and institutions which govern the the world of work. Now why isn't it – is it important? Well, AI is already shaping the labour markets, the world of work more broadly. And then with this report, the Director-General wanted to explore the kind of policies that we
need to have in mind and the kind of institutions in order to design and implement these policies. What kind of policies? Well, we need policies to facilitate the transition of workers whose jobs will be affected. Already even in in a scenario where there is not, you know jobs do not disappear at scale, we still know through the ILO research in particular that the effects of this transition will not be the same for all. You have some groups of workers
that will be more affected than others. So we have to design tailor-made policies in terms of facilitating the transition. Of course we need to adjust the skills and education systems in order to be future ready to respond to the new skills lens and I guess we will have also colleagues discussing about this particular issue. We need social protection measures to accompany those that are undertaking an upskilling and reskilling, let's say, process because it is
not – you know these people will not transit to another job from one day to another. They will have to have for instance some income protection in the meantime. And of course we need to support those enterprises and those countries that do not have access right now and they cannot benefit from this new technology which has a lot of positive aspects of course. And I refer mostly to the micro small and medium enterprises or to some developing countries that do not have access to this AI infrastructure. And of course we need to devise policies that,
I would say, democratize the use of AI in order to improve services, social services, labour inspection services or employment services. Yeah, so those are the key messages of the report. Thank you Kostas for setting out the framing of that moment of choice. Janine, now let's turn to the evidence base and the data. What do we actually know today about how AI is affecting jobs both in terms of quantity and quality? Right. So, our
research here at the ILO has been mostly focused on the potential exposure of certain occupations to artificial intelligence and what that could mean in terms of jobs, job losses, but also in terms of the transformation of people's day-to-day jobs. And so in essence it's actually a kind of a theoretical exercise because it's really about what could change and then there's of course what has changed what have we seen really in practice. So let me first talk about what could change. So our research shows that 20% of of occupations across the world are what we call exposed to AI
meaning that those people perform tasks that could potentially be performed by artificial intelligence. Now that doesn't necessarily mean and one shouldn't assume that that means that one fifth of jobs are going to disappear. It's not that at all. What it means is that in a particular job, we all do certain different tasks in our jobs, and some of those tasks could be automated. Now if the only tasks that one person does could be automated by that technology then the chances of that person losing their jobs is high. But if the person is doing a task where they do a
lot of different tasks and just some of those jobs could be done by the technology then there's more likelihood that that person's job would be what what economists call augmented but really what you mean productivity gains. So for example, if I want to translate something into another language and I use AI, I'm more productive because I have this technology that facilitates my translation. So, in short, one fifth of jobs that have the potential to be translated – the potential to be transformed or exposed and this varies depending on countries. So in the richer countries because
of the diversification of their economic structure you have almost a third of jobs that are exposed to the technology whereas in poor countries it's one in 10 jobs that are exposed to the technology. So that right there goes to what you're saying about how this transformation is not equal because in some cases you're going to have some countries that are not at all affected and on one hand that could be good because they won't have job losses but on the other hand that could be bad because they don't have the productivity gains. Okay. And then skipping to what has – what do we know in practice. So here the data is a lot more sketchy because it's hard to
identify what jobs really have been lost to AI. What we see instead is more of a slowdown in hiring in certain occupations. So you – we've heard reports for example of software developers, customer service representatives, people that could be replaced, for example, by chat bots. Those are some of the occupations where you hear that there actually have been some job losses. But what the more likely effect is is that there's going to be less hiring. And this of course has consequences for young people. It also has consequences for economy. So we need
to be doing more work. And that's one of the plans of the ILO and our Research department to be doing more research on actually seeing the real time effects of how this technology is affecting jobs. And I guess I've probably taken up my time so I'll pass the floor to Hannah. Thank you. That helps ground the discussion and what data is already showing. Now I'm turning to you Hannah. Kostas mentioned skills, and building on that let's look at skills and adaptation. What skills are becoming most important in AI-enabled labour markets and are institutions and
training systems keeping pace? Yeah, so at the ILO we just released our new flagship report on lifelong learning and skills for the future. And in this report we analyze online vacancy data to understand how skills demand is changing. And what we find generally speaking is that employers increasingly look for workers with rounded skills profiles. And these rounded skills profiles are associated with good working conditions. So a narrow focus on technical expertise alone
would be insufficient. And specifically in the context of AI, it's quite interesting that the overall demand for AI-specific skills is still comparatively low across countries. And we expect this low demand to still increase going forward. But we also think that it's due to the fact that many workers, they actually rely on ready-to-use AI tools and these do not require AI expertise
so much, but they require foundational skills like critical thinking or digital literacy. And related to this we also expect that social emotional skills will become even more important because these cannot be performed by machines. Now I think the second part of your question was whether training systems are keeping pace, and based on our new report we raised two concerns in this regard. So the first concern is that access to quality learning is still out of reach for many
adults and it is highly unequal. So, to give one example, only 16 per cent of all workers participate in training or structured learning over the course of a year. And many informal workers in particular, they learn mostly by doing without additional support or options. So that means that there's still a very long way to go before lifelong learning systems become inclusive.
And if I may, we raise an additional concern – this pertains to the specific role of social emotional skills, because as I mentioned, these are in high demand and we also find that they are associated with significant wage premier. But our research shows that this is actually not the case in the care economy. So for care economy workers, we do not find these wage premium. And this leads us then to conclude that yes obviously skilling, reskilling and upskilling are highly
important but it is also important that societies and markets adequately value the skills of all workers. Thank you. So we're seeing both opportunities and adjustments, pressures on the skill side. And Valerio, let me bring you in on the risks dimension. What are the biggest risks for workers emerging from AI today? And are current labour standards and institutions keeping up or do we need new approaches through social dialogue? Well, first of all, let me say that
this is a very important report that maps a lot of the existing questions that need to be answered at the moment. And so when it comes to the risks the workers face, again in many cases people tend to overfocus on the question of job quantity, how many jobs are going to be lost to AI, but I think that the report does a very good job of also putting the picture on the quality of the jobs that will remain and indeed there are some concerns that emerge when we talk about that.
And in many cases even when those concerns about job quality are voiced it's only about privacy and the question of data governance, which it's quite important but it's not the only point in the picture. And again the report does a good job of explaining what other risks workers face. First of all there's the question of how managerial prerogatives are expanded by AI and magnify the power and authority that exists in workplaces. And this increases the pace of work,
increases surveillance of work, which creates risk to occupational health and safety, which in many cases are not treated in the public sphere. Work becomes more intense and more demanding, and this intensification is driven by the technology. So this is one side of the picture that needs to be brought up. The other thing is non-discrimination, because these systems are mistakenly perceived to be neutral while they incorporate biases that exist in
society in the data sets and in the decision tree. So they reverberate biases and discrimination in society. There's risks to collective rights because some of this surveillance can and is being used to prevent people from unionizing or organizing in trade unions. And also at the same time, AI – and this is another thing that I found very interesting in the report that the report points out that in some cases the AI system that replace human decision making strip workers
and managers of authority and they actually, the workers need to re respond and answer to a sort of technocratic and technological authority in the workplace. So there is that dimension. And when it comes to the question of standards, ILO standards already do a very good job of establishing a framework of principles that can be followed, but there are some limitations, especially when it comes to the question of algorithmic management. And when it comes to
algorithmic management the ILO can do a very good job of incorporating social dialogue in the devising and production of these systems before they are introduced in the workplace. So this is where the key dimension but the tripartite dimension of the ILO can play a very good role. Thank you. I think after this first round we have a clearer picture of both the opportunities and the risks emerging from AI. Let's move on now to the diagnosis and to policy and responses. How can we ensure that AI-driven changes in the world of work lead to decent work outcomes across different
countries and groups? And I'm going to ask the same question to the four of you. I don't know who wants to go first. Maybe Kostas? Yes. Thank you. With pleasure. I will give a generic response. Okay. Which is a response that runs throughout all five chapters of the report. I think in one word we need good governance. Good governance will lead to decent work outcomes. And let me break it down into three components, this good governance. Let's say the three
dimensions, which we can identify throughout the report. The first is an active governance, meaning we know that in every technological disruption there are winners and losers. Okay. For instance, electrification and mechanization in the early phases of the industrial revolution led on the one hand to increased productivity and profits but at the same time to suboptimal working conditions, including let's say extreme cases of child labour or more intense working
conditions and working hours. It was only after appropriate, let's say, rules of the game were adopted, policies for fair competition, policies for – labour policies – that the costs and the benefits of technology and increased productivity were more fairly distributed, and the workers managed to profit. So that is the first component. The second component of good governance is inclusive governance and there I will agree with Valerio on the importance
of social dialogue. We know we have – I mean the DG says this in his report – but we know also from earlier technological disruptions that social dialogue is important. Social dialogue taking aboard workers' and employers' views in designing and deploying technology is important with a view to coming up with more legitimate, more workable, I would say, policies not only in order to protect the rights of workers but also to improve organizational performance.
So this is –and there is as the DG mentions in his report – there is clearly a business case for social dialogue especially in this area. And third the last component, last but not least, we need international governance. We need multilateralism. AI and the business models organized around this new technology do not recognize borders. No country alone can address,
let's say, its impacts positive or negative. And of course not all countries are faced with the same challenges. We know that there are, for instance, digital divides that need to be addressed. And this necessitates international cooperation. Without international cooperation some countries would not be able on the one hand to mitigate the negative impacts or on the other hand to profit from the advantages that this new technology brings about. And of course in multilateralism and international governance I include also the role of the ILO – the
ILO needs to be future-ready in order to be able to support the constituents in whatever policies they decide to design and implement at the national level. Thank you. The ILO needs to be future-ready. Who wants to come next? Maybe Janine or Hannah? Yeah. So we know for example – I mean as we've been talking about – that the effects of AI are not going to be equal across groups, across occupations, sectors, countries. So there's all these different breakdowns of how people are going
to be affected. And one of the real reason – I mean one of the big motivations for doing the ILO's numbers on occupational exposure is to give policy makers that information so that they can identify who are the groups of workers and what are the occupations that are most at risk. And with that type of information then you can develop proactive strategies rather than being reactive. And so what we want is really this proactive thinking by governments and social partners about how to respond. And so we can also think of, you know, yes we need to have more, you know, if there are going to be some job losses, and there might be, what could be some new jobs
that be created? What are the sectors where we know that we need to be investing and could be a source of employment? We know that there's shortages in the care economy across the world, poor countries and rich countries. We know that there is a need for the green transition and a lot of work in that area. So if you start thinking about developing, you know, shifting some of these occupations and some of the investments in countries to these areas, this could be a source of job creation. And of course, people need to be supported during this transition. How can we have social protection systems that are robust enough to support workers during that transition?
People need some sort of income security when they're doing training programmes, for example. So that's one thing. The other issue is about the job quality aspect. So as I mentioned, a lot of jobs are going to be transformed, not necess – you know, so they might not, you know, they won't be necessarily lost, but they could be transformed. And this could be positive or this could be negative. And this is where this whole issue of governance and choice is so important, because for that transformation to be possible, it's really important to have a process that is participatory , that does use social dialogue so that you can get workers' perspectives along with the employers'
in how the technology can be integrated into the workplace. Workers know their jobs best and so when technology is adapted with the workers in a participatory design process, you're more likely to have technology that is more productive, you're more likely to have a situation where working conditions are improved. So, those are some of the things that can be done. And we know also that, you know, some of, in some countries with stronger industrial relations systems and stronger collective bargaining, a lot of collective bargaining agreements have come out in the past few years where there has been explicit safeguards and involvement of workers
in that process that can be beneficial. But I mean I think the real message is this message about being proactive and not being reactive, and thinking, you know, what can we be doing to help manage this transition. Thank you. Hannah, you want to come in on that – proactive? On being proactive? Yes. I think that may be one of the conclusions of this whole discussion, no? But I will respond again from the perspective of lifelong learning, given this new flagship report that we just published. And because in the era of AI, it's obviously important that workers
can acquire new skills. And this needs to be true for workers of all ages and also for workers with different levels of formal qualification. And based on our research, we argue that this really requires an ambitious systemic approach, and that has several implications. So first of all, it requires acknowledging the many different ways in which people learn, which clearly go beyond the traditional classroom, to make sure that workers' skills are recognized and also that they have
access to adequate qualifications. The second implication of this ambitious systemic approach is more conceptual in nature. So lifelong learning systems should be designed not only with narrow objectives in mind like employability or productivity, but they can also be used in a way to advance broader objectives like decent work through innovation, active citizenship or social
cohesion. And then finally more from the perspective of policy, we define the building blocks of successful lifelong learning systems. So financing is one of the elephants in the room. We need sustainable financial solutions that are understood as a shared responsibility. And then related to what Kostas and Janine said, we obviously also need a strong social dialogue. And Valerio? Yeah, I think that one of the things that need to be faced is that
at the moment workers are not involved at all in the design and the production, let alone the introduction of the systems that actually are used in the workplace by them or over them. And this creates, as Janine said, many shortcomings in how these systems operate. We know that most of AI pilots in businesses fail after a short implementation, precisely because they
are introduced top down on workers without any feedback from them. And actually workers are the ones that know how the job is actually done. So there is a lack of involvement that also is a waste of time for employers and a drain of resources for employers. So what we actually need is fostering the dialogue between capital and labour, employers and workers to come together and have a say in the technology that is implemented in workplaces.
At the moment, most of these things are decided by tech companies and tech people who have absolutely zero idea of how a job is done and how it actually works. And this is what is creating a lot of waste for everyone and negative externalities for societies. Thank you. And now we're going to move to our final – our third and final round and I'd like to ask each of you to look ahead. I'm also going to ask you, if possible, you can stick to one minute for this final message. And the
question is, what is the most important priority to ensure that a human-centred future of work remains achievable in the age of AI? I don't know who wants to start first, but please go ahead and final message in one minute if possible. Okay. So I think it should be strengthening social dialogue and collective bargaining over the design and introduction of algorithmic management at work, again for the benefits of workers, employers and societies alike. It's really tricky, because
a hundred different aspects come to mind, and I think the DG's report to the ILC picks up on many of them. But I will select a point that Janine already alluded to. So there's now a debate whether entry-level hiring has stalled in certain white collar occupations. We still need more evidence to be able to answer that question conclusively. But what's sometimes missing in the debate for me is the realization that the entry- level workers of today will be the mid-level and higher- level professionals of the future. And so I think it can be very much in the interest of enterprises
themselves to have a longer-term planning horizon and to invest in this human potential. So I actually don't think there's one important priority. I mean if I had to pick one and maybe it would probably be the dialogue but but it's just like the the technology itself. People think it's easy to use. It's shiny. It works well. And so you you put it in simple solutions, you'll get great results. And it doesn't work that way. And it's the same with policy-making as well. You have to have a lot of different approaches to achieve the ultimate objective
of improving well-being, you know, improving human welfare. And so that means you need to be, you need to be investing resources in job creation. You need to be investing in improving working conditions. You need to be investing in in reducing the digital divide. There are so many different fronts that need to be tackled, which is why I think governments and social partners need to be really working together to develop comprehensive plans and enacting them, and of course dealing with the financing issues. But if I had to sum it up in one I would say dialogue between all these different parties and that recognition that yes it's going to be a hard –
it's a long haul and it's a hard haul and we have to go forward with it. Yes. Thank you. Well, from a governance perspective, first of all, I agree it's very difficult to choose a priority, but I would say from a governance perspective, I think the priority is twofold, composed of two interrelated elements. The first is having a solid knowledge basis on the benefits and on the risks of this new technology. Obtaining and sharing this information
on the risks and opportunities will be extremely important for designing and implementing policies, the policies which we mentioned before. So we need to identify both the opportunities and the risks, and we need to do it all the time, given the pace of developments and evolution of this new technology. So – and the second component, which is related to the first one, always again involve those directly affected by these decisions, by these policies,
workers and employers. And I would say even addressing, inviting also the experts to talk to the workers and the employers. Because in addition to the social partners participating there, they have to have the capacity not only they have to be invited but they have to have the capacity to participate effectively in these policy-making processes. So I would say those are the two most important components. And if we have both information and the capacity to participate
effectively, I believe both the legitimacy and the efficiency gains that we can win are opening very positive and new horizons. Well thank you very much to the four of you for this extremely rich discussion. If there is one message from today's conversation, I think it's that the impact of AI will depend on the choices we make today through inclusive dialogue. Thank you for joining us at the International Labour Conference in Geneva. And be sure to
tune in again as we continue to explore the major transformations shaping the world of work today. You can follow us online @ILO on X, Tik Tok, YouTube, @ILO.org on Facebook, Instagram and Threads and Blue Sky and also @InternationalLabourOrganization on LinkedIn. Thank you for listening and we look forward to welcome you again soon. Goodbye.
