The future of jobs in an age of AI and automation has been a popular topic on this show. We’ve tended to talk in quite theoretical terms and never truly considered what practical solutions there might be to the inevitable job displacement. Millions of jobs have already been lost in the pandemic, and the World Economic Forum predicts that an additional 85 million jobs with be lost to automation by 2025. So what type of jobs will disappear, and how do we upskill and transition the large proportion of the workforce who will be affected?
My guests this week are Madeleine Gabriel and Karlis Kanders from Nesta, the UK’s innovation agency for social good. Nesta has just completed a significant piece of research called Mapping Careers Causeways, which is intended to support job transitions and inform skills policy in a labour market that AI and Automation will forever change.
In the interview, we discuss:
▪ The uncertain future of work
▪ Which jobs are at risk of automation
▪ Identifying pathways from high-risk occupations to safer ones
▪ Why women are twice as likely to be displaced than men
▪ The three types of core skills that will help people transition
▪ The critical importance of training
▪ The role of talent acquisition
▪ Thinking differently about hiring
▪ Why the best way to predict the future is to create it
Listen to this podcast in Apple Podcasts.
Transcript:
Matt Alder [00:00:00]:
Support for this podcast is provided by SHL. From talent acquisition through to talent management, SHL’s science and technology maximize the potential of your greatest asset your people. SHL help you create the diverse, agile and innovative workforce you need to succeed in an unpredictable environment. Their data driven people insights, unmatched portfolio of products, engaging experiences built on science and global expert services are all delivered on one platform for all your people. Answers Visit shl.com to learn more about how SHL can unlock the potential of your workforce.
Matt Alder [00:01:06]:
This is Matt Alder. Welcome to episode 336 of the Recruiting Future podcast. The future of jobs in an age of AI and automation has been a popular topic on the show. However, we’ve tended to talk in quite theoretical terms and never truly considered what practical solutions there might be to the inevitable displacement of job. Millions of jobs have already been lost in the pandemic and the World Economic Forum predicts that an additional 85 million jobs will be lost to automation by 2025. So what types of job will disappear and how do we upskill and transition the large proportion of the workforce who will be affected? My guests this week are Madeleine Gabriel and Karlis Kanders from Nesta, the UK’s innovation agenc for social good. NESTA has just completed a significant piece of research called Mapping Career Causeways, which is intended to support job transitions and inform skills policy in a labor market that will be forever changed by AI and automation. Hi Madeline, hi Karlis, welcome to the podcast. Please could you introduce yourselves and tell us what you do?
Madeline Gabriel [00:02:32]:
Hi Matt, thanks for having us on the podcast. I’m Madeline Gabriel, I’m a social researcher by Background and now I’m a Deputy Director at Nesta and I focus on economic recovery and sustainability projects.
Karlis Kanders [00:02:43]:
Hey Matt, thanks for having us. Also it is really great to be here and actually I’m quite excited because this is my very first time being on a podcast. So I’m a data scientist at NESTA and I worked in the Mapping Career Causeways project that we will talk about today. And previously I completed a PhD in computational neuroscience at the ETH Turk in Switzerland and there I studied the structure and dynamics of complex networks, for example real developing neural networks grown in laboratory. But now for the past year and a half I’m studying different kinds of networks, namely networks that consist of jobs and skills. And so during this Mapping Career Causeways project I was looking at how workers can move in these networks and find better jobs.
Matt Alder [00:03:26]:
Fantastic. And we’re going to dive into that in a bit of detail in a second. Before we do, though, it’s probably worth just giving people a quick overview of NESTA and what NESTA does in case people aren’t familiar with your work.
Madeline Gabriel [00:03:38]:
Yeah, sure, I’ll do that. So we’re at Independent Charitable Foundation. We were set up about 20 years ago and our purpose is to design, test and scale new solutions to society’s biggest problems. So, in other words, we often talk about innovation for social good. One area that we’ve been working on for the last few years is focusing on helping workers adapt to a changing labour market and avoid unemployment. And we’ve been doing quite a lot of different things in this area. One of the things we do is fund and support innovators. So we’ve got a program, for example, called Career Tech Challenge, where we’ve got 20 innovators who are developing new ways to provide information, advice and guidance to adults. We’ve got another one called Rapid Recovery Challenge, that is about trying to connect younger workers who’ve recently lost their jobs, maybe because of COVID into other jobs that match their skills. And then a big area of our work is doing our own research and development and mapping Career for Causeways, I think, is one of our most exciting projects and that’s why we’re keen to chat to you about it today.
Matt Alder [00:04:38]:
Absolutely. And, you know, it’s a. It’s a phenomenal piece of work and it’s something that I know everyone’s going to find really interesting. I suppose just to set a bit of context before we get into the detail, one of the things that you’ve been looking very carefully at is the labor market trends that are going to shape the future of work over the next ten years. Could you. Could you talk us through what those key trends are likely to be?
Madeline Gabriel [00:05:03]:
Yeah, we did a big piece of work on this back in 2017, where we looked at seven major trends that will shape the future of work. I’ll say right up front, the one that we didn’t look at was global pandemic, so we didn’t miss the boat a little bit. But there are other things that are really going to shape the future of work. So one of them, of course, is automation. And that’s quite a factor that we look at in detail in mapping careers causeways. But there are others too. So we think, for example, sustainability transitions are going to shape opportunities. So a shift away from highly polluting Industries, for example, towards greener jobs. Demographics obviously makes a difference in aging population in a lot of richer countries. There are others around things like social inequality and what difference that that might make if it continues to grow. Urbanizations and more people obviously living in cities and towns. Globalization, which, you know, I think when we did the work in 2017, the assumption was that globalization would continue. But we’ve seen some retraction of that in the last couple of years. So there are quite a lot of changes. And I guess one of the key messages of that work is that there’s a huge amount of uncertainty. So that work tried to look at which, which jobs and occupations are likely to change, which ones there’ll be fewer of and which ones there’ll be more of. We’re able to predict contractions in some areas and growth in others. But for the vast majority of workers, the future is actually quite uncertain. And I think that’s, that’s a key message really.
Matt Alder [00:06:29]:
And that obviously leaves us, leads us into the Mapping Career Causeways project that you’ve, that you’ve done. Why did you do the research and how did you do the, and what’s the research about?
Karlis Kanders [00:06:40]:
In a nutshell, the Mapping Career Causes project, which was funded by JP Morgan as part of their new Skills at Work initiative, presents a new framework for supporting job transitions and also informing skills policy in a changing, uncertain labor market. And so importantly, this framework, which is essentially an open source career transition recommendation, so it’s fully open and transparent and it can provide very granular recommendations for over 1600 different occupations. So the initial key aim of this project was to support workers whose jobs are at risk of being automated by machine learning algorithms. And specifically we were interested in identifying how these workers can build on their existing skills and experience and ultimately transition to more secure jobs that are less likely to experience technological disruption. And for the context, this work started before the Coronavirus pandemic. And at that point the major disruption we anticipated was automation. But partially in response to the ongoing pandemic, a second aim emerged which was to develop more general methodology and sort of a data driven toolkit for measuring the resilience of workers to various types of economic shocks, for example, including COVID 19. And so this is the why of the project and then how did we do it? And I’ll break it down kind of into three parts where the first part was to estimate the risk of automation for these 1600 European occupations. And then the second part was to look at the actual workers whose jobs are at risk. And so we did this by identifying demographic patterns using the European Union Labor Force survey. And we did this for three countries, for the uk, For France, and for Italy. And then thirdly and importantly, we essentially built a map of occupations that can help to identify potential pathways from these high risk jobs to safer occupations. And so to achieve this, we brought together multiple data sets, including the official European and American occupational frameworks, labor force statistics, and these automation risk estimates that we made. And the crucial step here was to measure the similarities between the different occupations. And so we did this based on their skills on the work activities, and also more broader interpersonal or physical and structural work sort of context features, for example, how long you need to talk on the phone, or how much face to face discussions do you have at work, or whether you need to stand or sit or do the work outside or inside and stuff like that. And so to compare these features, we used methods from data science and machine learning, for example, natural language processing, to compare different skill sets and find the transferable skills as well as the largest skill gaps for any career transition. And this is not all. So to complement this quantitative analysis, we also did some more research and workshops and interviews and crowdsourcing, so sort of more qualitative work. And we engaged local stakeholders, for example, learning providers and public employment services in the uk, France and Italy, to better understand their needs and challenges. And my colleagues interviewed more than 50 different organizations to understand how these insights could complement their work and what would be necessary to take this work forward. And finally, and sort of last but not least, we also recently performed a crowdsourcing experiment to validate the career transition recommendations by our algorithm. So we kind of asked a crowd of people to rate around 10,000 of the transitions and kind of give a common sense judgment whether they think they are feasible or not. Then we use these judgments to refine our recommendation algorithm.
Matt Alder [00:10:31]:
So to bring that to life a bit for the people who are listening, could you sort of talk a bit about the results or maybe even give us an example of a job that’s going to be automated and the career opportunities that might be open to people transitioning out of that kind of role?
Karlis Kanders [00:10:46]:
Yeah, so first of all, who is at risk? So we find it maybe unsurprisingly, sort of workers who are working in predictable environments and doing routine actions, including also one way routine interactions with other people are mostly at risk. So in terms of sort of more specific occupations, these would be jobs in the retail and customer services or clerical business administration roles that would be impacted the most and to some extent Also finance related professionals. So some maybe specific examples that I can name from top of my head would be like cashiers or sellers or file clerks for example, or investment clerks, or even something like property appraiser or surveyors. And I should also definitely mention that by looking at the demographic patterns and looking at the EU labor first survey, we unfortunately find that actually women are twice as likely to be in these high risk occupations than men. And this was true across all three countries that we studied, like the uk, France and Italy. And also importantly that at risk workers are likely to be lower paid. And therefore it could be that those at the highest risk of job displacement may also be the ones with fewer financial resources to weather the disruption. And of course with all this I should kind of also flag a disclaimer that kind of these results are sort of the potential theoretical impact of automation and that the actual adoption of automation is a different thing. And actually we do lack sort of, we have a lack of good data on the actual adoption of automation or machine learning solutions at the moment. So it’s important to keep in mind that these are theoretical estimates. But kind of the next question that I think we also been asked or should ask is where, given that there’s a potential risk for many workers, so several million in each of these countries, what are their transition options to more secure jobs? And so one sort of the first kind of striking finding that we found was that kind of jobs that require similar skills also have similar levels of automation risk. And so when we visualize this map of occupations that is also published on Estes website, that everybody can also see for themselves and explore it, you can clearly see sort of these red high risk areas in the map. And what this means is that the viable transitions for risk workers are not necessarily automation safe. And so this is when we compare the high risk occupations with the lower risk occupations, we find that the high risk ones have 40% less transition options that workers in the lower risk jobs. So this tells us two things, two important things, that it’s important to have this information. First of all, otherwise we’re kind of navigating around in the labor market blindly and not necessarily making sustainable career choices. And secondly, that at risk workers might have to make larger leaps to reach safer occupations. And so there has to be support systems in place to guide and facilitate retraining and upskilling. And this brings me to the kind of last point on this segment that so what, what are the sort of most promising retraining or upskilling opportunities that could open up new and safe career opportunities. And so first of all, our research kind of confirms that training is important. Like using our model, we find that workers with higher levels of training have more transition options. And this training can be of different kinds. It can be on the job or it can also be through education. And now in terms of exactly what kind of training can help, we can also use our algorithm to kind of tease apart this question. And so we find that there are three major types of core skills that on average unlock the highest number of new transition options for any at risk worker. These three types are management skills, communication skills, and information analysis and evaluation skills. These core skills kind of emphasize the potential of non routine activities that require advanced cognitive reasoning, human judgment and working with people to protect workers against automation. And so the opportunities for acquiring these kind of skills should be actively pursued, right, and supported. And these core skills that we haven’t covered can be actually naturally gained on the job if more delegation and independent decision making would be encouraged. Or alternatively, they could also be acquired perhaps through more informal routes like for example, volunteering.
Matt Alder [00:15:14]:
This is a very informative and, you know, very, very interesting and slightly alarming piece of research in terms of actually mapping this situation. What are the intentions, your intentions of how the research should be used, who should be accessing it, and you know, what role might the talent acquisition community play in all of this?
Madeline Gabriel [00:15:37]:
Yeah, we see probably three main use cases, although we’ve been having a lot of discussions with people about it over the last couple of months, and new ideas keep coming up. But the three main ones are probably first and foremost supporting career moves. So helping people identify transitions that are viable for them based on the skills they’ve got and also that are desirable. So based on whether the new job is at the same salary or more to their previous role. And as Carlis said, it can also identify which occupations are safer from automation. So I suppose if you lost your job or are just looking to move careers into something that’s, you know, better, better suited or more interesting or, you know, more sustainable in the longer term, you could think about what those things are. And we kind of assume that when people are thinking about career changing or perhaps if they’ve been made redundant, they don’t necessarily have a very broad idea of what’s available in the job market. What we hear from people is that who are working in this space is that sometimes people’s horizons are quite narrow. And if I think about my own career, I don’t really know much outside of the charity sector. So I think one of things that we think this could do is help broaden people’s horizons and suggest options to them they might not have otherwise thought of. So that’s one use case. Another, as Carlis was just saying, is around providing upskilling and retraining advice. So, for example, if you’re in a job currently and you know what job you want to move to, this model could help you identify which skills you probably have, which skills you might need but probably don’t have, and help you think about what, what training you might need to do in order to access that job that you want. Another potential use for it is where you actually don’t know what jobs you should move into, but you know you can’t stay in the thing that you’re doing. Perhaps there just aren’t enough opportunities or you just want to do something different. In that case, based on the data we have, we could recommend what new types of training would open up the greatest number of opportunities for you. So which additional skills might help you do the most new things. So one example could be like, if you’re a shop assistant, this model might recommend that you learn management skills because that would open up quite a few more transition options for you than you might otherwise have. And then the third one is a bit more strategic. So thinking about how to support workers at risk, so you could map the impact of shocks, as we’ve done with automation, but you could overlay other types of shock onto the map. So the impact of COVID or the impact of green transition, et cetera, and then understand which groups of workers are going to find it most difficult to kind of escape that shock. And therefore I would most likely need additional support. And we think there’s quite a few different groups who could make use of this. So we’ve been talking to providers of careers advice and employment support. Of course, that’s a big growing sector at the moment with the impact of COVID on redundancies in the labour market generally. We’ve been talking to local authorities and others who are working in policy and strategy and also I think employers themselves can make use of it. So there’s some, say large organizations who want to think about how they can apply it for learning and development within their own organisations or maybe think about how they internally redeploy staff. And I think there’s potentially other uses as well. I mean, we’re quite interested in talking to recruiters about how far employment practice or recruitment practice lends itself to a skills based model like this. So basically, would recruiters consider. Consider Applicants who’ve got their experience in different sectors and maybe look a bit different from people who are normally employed in those companies. And I’d be really interested in your perspective on that and also what your listeners think of that.
Matt Alder [00:19:28]:
Absolutely. And I’m sure there are lots of people listening who want to find out more and would love to have a conversation to see how that could develop basically. I suppose with that in mind, it would be a good sort of time to remind people where they can actually find the research. So you mentioned it was on your website. Could you just sort of, you know, where, where can people go to to find out more?
Madeline Gabriel [00:19:48]:
Yeah. So our website is Nesta, that’s N-E-T a.org UK and if you were to search for Nesta mapping career causeways, you’d get straight to, to the report. What we’ve got on online at the moment is the quite a detailed report that talks through all of the methodology and research findings that Karlis talked about. We’ve also released the code and the data on GitHub, so if you’re a data scientist, you can go there, use this stuff yourself. And shortly we’re going to be putting out user guides that are aimed at those core audiences and just trying to show what the use cases of this work and how they can access it. So that’s, I think, going to be released in March. Is that right, Karlis?
Karlis Kanders [00:20:29]:
Exactly. Yep.
Matt Alder [00:20:31]:
So as a final question, obviously we talked right at the beginning about the pandemic has made us realize that making predictions is not a science because we don’t really know what’s going to happen next. But based on the research that you’ve done and the trends that you’re, you’re looking at, what do you think the, the future and jobs, the future of jobs and works looks like, you know, over, over the next few years, are we going to see mass unemployment? Are we going to see new jobs and new careers being invented? What, what’s your kind of perspective on where we might be heading?
Karlis Kanders [00:21:05]:
So, as Melin explained a bit earlier, you know, the future is really uncertain and there is a whole host of trends that could affect it. But I wanted to give an optimistic prediction and it’s more to do with the kind of future career technologies that we will have. And I personally see a future where employment services and workers will have at their disposal amazing and advanced career planning tools. So be it websites or apps on your phone, when I think about these tools, I think it’s something like Google Maps for jobs and skills, you know, like Google Maps allows you to locate yourself in the world and find what’s around you, be the cafeteria or pharmacy or cinema. And it allows you to find a path right from point A to point B. In the same fashion, I think we’ll have career planning tools that will locate you in the occupational landscape based on your competencies and experience and will show you the other jobs where you can use your skills. And moreover, it should also show you how to get from where you currently are to where you want to be. And it should point out also the skills that you need to learn on the way and where you can learn them and how much time and money it will take. Now, kind of in this brave new world of algorithmic career guidance technology, I would also kind of like to give a few words of caution or things that we should keep in mind. And my colleagues recently came up with four key principles that they formulated how these new career guidance tools should work. And let me just kind of maybe tell you these four principles. So the first one is that they should be viewed as complementary rather than a substitute to the existing tools, advice or information sources. Secondly, they should be open to scrutiny, so completely transparent, so that if there is any bias or any potential kind of, you know, misinformation there, then it should be transparent enough so that it’s easily identifiable. Thirdly, these kind of tools should only be used to broaden options. So we don’t want to limit people’s choices and push them in some career that maybe is not best for them. And finally, the advice should be focused, of course in the long term, so providing sustainable recommendations for jobs that will remain safe from automation also in the coming years. So kind of to bring it together. I would say, you know, as the saying goes, that the best way to predict the future is to create a. And you know, mapping career crossways projects makes a humble step in this optimistic direction. But I think as another example, we can also get a glimpse of this future by looking at what also Madeline mentioned in the beginning, the Nestus Career Tech Challenge, which is this competition where there are several about 20 companies that are competing to create digital solutions and improve the access to accurate career advice and guidance. And this competition will close this year in March when the winner will be announced.
Matt Alder [00:23:57]:
Fantastic. Madeleine and Karlis, thank you very much for joining me.
Madeline Gabriel [00:24:00]:
Thank you Matt.
Karlis Kanders [00:24:01]:
Thank you very much, Matt.
Matt Alder [00:24:03]:
My thanks to Madeline and Karlis. You can subscribe to this podcast in Apple Podcasts on Spotify or via your podcasting app of choice. Please also follow the show on Instagram you can find us by searching for Recruiting Future. You can search all the past episodes@recruiting future.com and on that site, you can also subscribe to the mailing list to get the inside track about everything that’s coming up on the show. Thanks very much for listening. I’ll be back next time, and I hope you’ll join me.
