AI, Employment, and the Opportunity Gap

Karen Bonadio:
Good day everyone. I am Karen Bonadio, director of alumni relations, and I'm delighted to welcome you to our last Alumni Talk Policy Zoom webinar of the semester on AI, employment and the opportunity gap. The Alumni Talk Policy series features HKS alumni and panel discussions about pressing public issues. While we cannot meet in person, technology allows us to convene virtually, and we appreciate your patience as we navigate this event remotely. This webinar is being recorded and closed captioning is available and can be turned on at the bottom of your screen. Today, I'm happy to introduce Chris Riback, MPP 1992, founder of Good Guys Media Ventures, who will moderate this panel discussion. Chris, I'll now turn it over to you to kick off today's important and timely discussion.
Chris Riback:
Thank you, Karen. And thank you everyone for joining, and our extraordinary panelists whom I will introduce in a moment. So Karen and panelists, as you all know, among the most urgent global challenges today is the series of seemingly unreconcilable gaps in wealth, opportunity and equity. These gaps are fueled by a myriad overlapping vicious circles, few more challenging than those around skills and employment.
Chris Riback:
The UN Future of Jobs 2020 report highlighted the double whammy on the one hand, automation destroy jobs, "By 2025, automation and a new division of labor between humans and machines will disrupt 85 million jobs globally in medium and large businesses." On the other hand, automation simultaneously accelerates new jobs. The UN estimates 97 million new emerging roles, including in the fourth industrial revolution technology industries, like our topic today, artificial intelligence.
Chris Riback:
However, for workers to keep their jobs, those jobs, the UN reports in the next five years nearly 50% will need re-skilling for their core skills, which is why we are having a discussion today on AI, business and policy, which seeks to address several of the central public and private sector questions around AI and the future of employment. And you will very soon understand why we have these three guests with us today.
Chris Riback:
Simon Mueller, an MC/MPA 2015, is co-founder and vice president of The Future Society. Shireen Santosham is an MPP 2009, and she is head of strategic initiatives at Plenty. Previously, she was Chief Innovation Officer for the mayor of San Jose. And Anna Stansbury, an MPP 2015, an economics PhD from just this year, congratulations, joins MIT Sloan this fall as an assistant professor in Work and Organization Studies. If she's going to MIT Sloan, then I guess we're going to have to work hard to win her back.
Chris Riback:
Simon, Shireen, Anna, thank you for joining. If you could maybe say a couple more words about each of you, we will do it in alphabetical order, by last name, which means Simon, you are first. And Simon, please feel free to give a quick public service announcement, since we are in the public service business, for The Harvard Kennedy School Emerging Technology Alumni Association, which can be found at HarvardTechAlumni.org. Simon, please tell us about you.
Simon Mueller:
Thank you very much, Chris, and thank you for having me on this panel. I'm indeed the co-founder of The Future Society, which is an organization, a student's organization that we founded 2014 at the Kennedy School, and we are now incorporated both in the US and in Europe in Estonia. Our mission is the governance of emerging technologies, specifically AI, and we work with a variety of organizations from the OECD to the World Bank to the GIZ to Microsoft, The World Government Summit, etc.
Simon Mueller:
With a couple of highlights, we have organized The Global Roundtable on The Governance on AI in Dubai as part of The World Government Summit, The AI Athens Roundtable on AI and The Rule of Law, and then for example, The AI National Strategy for the country of Rwanda last year. The Harvard K School Emerging Technologies Alumni Association is in fact a SIG, special interest group, and we welcome anyone to sign up for events. If you are interested in the questions of impact and governance of emerging technologies such as AI, but not limited to it, please do join us, HarvardTechAlumni.org is the address, and I'll also send that in the chat. Oh, thank you, Rachel.
Chris Riback:
Thank you, Simon. Shireen, could you tell us a little bit more about you?
Shireen Santosham:
Hi, everyone, thank you for having me on the talk today. I would be remiss if I didn't say that I was actually a MPA/ID '09 and I joined you at the business school. And so, I currently work at Plenty, which is a high tech, indoor farm, we use advanced robotics AI to disrupt, essentially the food system, and so our firms can go anywhere, in the middle of a city. We condense hundreds of acres of farmland into the size of a big box retail store using a fraction of the water, and so we're sort of building these farms at scale. We currently have one in San Francisco, and are building the largest vertical farm in the world in Compton this year. And so, it's a really exciting space, and really combines a lot of the topics that we're going to be discussing today.
Shireen Santosham:
Prior to this, I was working as the Chief Innovation Officer for the mayor of San Jose, which is the other big city outside of San Francisco in Silicon Valley, and so home to many of the major tech companies, including Zoom's headquarters is in San Jose. And so, worked on tech policy at the local level, but also with the largest companies in the valley, as well as interacting with the federal leaders, mostly because they of course come to the valley to understand how this technology works. And so, really excited to join this panel, and I'll turn it over to Anna.
Anna Stansbury:
Thanks. Thanks everyone for having me here, I'm very, very excited to be here and have this conversation. I'm a labor economist just as Chris noted, just finished my economics PhD at Harvard this year and starting at MIT as Professor in the fall. And my work focuses on how we can structure labor markets in industrialized economies like the US to provide good jobs and shared prosperity as much as possible.
Anna Stansbury:
And so I think a lot about the questions of how we should respond to emerging technologies in the future of work, and I think a lot about questions around power and institutions in labor markets. Who has the power to set wages? Who has the power to determine working conditions, and how government organizations like unions and business can respond to develop working conditions and good jobs and reduce income inequality?
Chris Riback:
Thank you. Thanks Anna, thank you all and let's dig right in. I'd like go right into that intersection and tension between business and public policy. It's where all of us have studied and worked, and where so many of the people on this Zoom right now work and think about. Given that ongoing and widening disruption and the role that AI will play in reducing existing tasks, creating new jobs, and requiring ever improving human skills to keep pace, what should be government and businesses roles to secure the best of both worlds? How should we think about that tension? Anna, why don't we start with you? How do you think about that tension from where you sit?
Anna Stansbury:
Yeah, so I think a useful kind of starting point here is actually just to start with a little parable of the Luddites, because I think this is where our modern narrative often gets a bit lost. So the Luddites are famous today for smashing machines in protest against new technology, and their name has become synonymous with the desire to slow technological progress. But it's less known that actually the Luddites were largely protesting against the destruction of their livelihoods by the technology and not the technology itself.
Anna Stansbury:
The Luddites were skilled textile workers in England at the beginning of the 19th century, owners of workshops whose workshops had closed, because factories using new weaving machines could produce cloth more cheaply. Their training, many years of training had been rendered completely obsolete, their livelihoods had been lost, and in an era of high unemployment, to add insult to injury, many of them couldn't even find new jobs for low wages in the factories that were often employing children and very poorly trained workers to weave cloth.
Anna Stansbury:
And it was only after many years of repeated petitions to Parliament for minimum wages, for protections of apprenticeships and skilled workers that the Luddites resorted to smashing machines. In return, the government made machines smashing a capital offense, sent the troops onto the protesters and executed many of the protesters for this crime.
Anna Stansbury:
I think this is important to think about, because when we think about technology today, there's often a narrative of slowing the pace of technology in the workplace, but obviously, our discussion today is instead about figuring out who it impacts and how. The parable of the Luddites is important, because it illustrates just how disruptive new technologies can be in the transitional sense, but it also illustrates how important it is that the power structures we have in society respond to allow the people who are impacted to have a say in how those technologies are implemented.
Anna Stansbury:
And I think it's easy to take a 30,000 foot view often in policy and say, "What do we think the ideal world would look like?" What we need to remember is, what do the people who are directly impacted want the world to look like? How do they want the technology to come into the workplace and how can we structure systems such that they have a say, a democratic say in the implementation of that technology?
Anna Stansbury:
I'll just say one more thing, I don't want to take up too much time, but I think another interesting distinction is to think about transitional dynamics versus equilibrium dynamics. So the transitional dynamics are, when we have a new technology that comes in, whether it's a loom in the late 19th century, or a self driving car, what happens to the workers whose jobs are directly destroyed, and whose livelihoods are directly destroyed?
Anna Stansbury:
Whose responsibility should it be to ensure that those workers retain a good living standard and can do a different job with dignity and with good pay for themselves and their families? Government, business, society. But the other is, once that transition has happened and once that shakeout has happened, what is the equilibrium labor market outcome that we will reach? Will the new jobs that have been created actually be better than the old jobs that have been destroyed? And in the Industrial Revolution, for the first couple of decades that wasn't the case. And one might also think that that's not going to be the case today.
Anna Stansbury:
Is it worse to be a Lyft driver than it was to be a taxi driver? Arguably, yes. Is it worse to be a minimum wage home health aide than it was to be a middle income manufacturing worker? Almost certainly based on the pay and working conditions that we have socially determined for those jobs. And so I think we need to bear in mind both of those different factors.
Chris Riback:
Shireen, obviously Anna was referring to me and not you with the Luddite comments, but the whole rest of what she had to say about the balance between the democratization of work, the opportunities, the sense of labor, having some say in how technology affects their output, their labor, their opportunities, these are areas that you have thought about, I would think, and are working in now. So in thinking about what you do at Plenty, in thinking about the jobs that are getting created in that environment, but also keeping in mind the work that you did with the City of San Jose, is it possible to extend the economic opportunity that AI promises without also expanding the opportunity gap that it threatens?
Shireen Santosham:
I think the answer to that is of course nuanced, you'll see... At the end of the day, I think Anna is absolutely right, that jobs are dignity, jobs are... Livelihoods can't be ignored, and sometimes when you talk about these very abstract concepts, whether it's AI or climate change and you ignore the impact on jobs, you're really not going to be successful in an outcome from a public policy point of view.
Shireen Santosham:
In terms of we'll take Plenty as an example our vertical farming. What is happening globally is we're running out of farmland. I mean literally, without chopping down the Amazon, or really going after some of our sort of sacred natural resources, we won't have enough farmland to feed our population, the world's going to need three times more fresh fruits and vegetables than we have today to feed everyone a nutritious diet. And at the same time you see, in farming, really certain jobs, for example picking strawberries are really backbreaking and labor intensive, and people don't want to do it, regardless of where that labor is coming from, if it's immigrant labor or local labor.
Shireen Santosham:
And so something like Plenty can actually bring jobs into a city, provide high tech manufacturing jobs and feed a population without necessarily "displacing workers." I think it's a sort of false narrative that you have to put these two types of farming in opposition to each other, when in fact they're quite complimentary.
Shireen Santosham:
And so I think that's a good example of where we can expand the pool. And in our case we're going into Compton which is a community that has a per capita income of about $17,000 a year and so we're providing good high tech jobs. That all being said, as a Chief Innovation Officer in San Jose, we'd often talk about San Jose as the tale of two cities. So you have a really high income tech workforce, or people who say bought their homes 30 years ago and their property values have just gone through the roof, and so are sitting on a lot of wealth.
Shireen Santosham:
And then you have a lot of low income families, like San Jose is over 30% Latinx, living in a two bedroom apartment with eight people with no digital access, right? And so that San Jose had gone from being one of the most socially mobile cities in the country to really starting to lose that because of the impact of the tech bubble. And so there aren't easy answers to it, but certainly is something that I think the local leadership in the valley is very aware of, and the tech world is increasingly waking up to.
Chris Riback:
Yeah, you are kind of at ground zero there in terms of the impact that technology that AI can have on a society in different towns and the urban environment. Simon, as you are listening to Shireen and to Anna, I know as well that something that you think about quite often is an environment where we are managed by bots, where the AI becomes part of not only what transforms labor and the worker level, but also the management layer. Any thoughts, comments to what you've heard so far from Shireen and Anna? And maybe you can talk about the thinking that you do around our all working for bots.
Simon Mueller:
Yes, yes. And let me actually start by saying, I think in your opening remarks, Chris, you mentioned The UN Future of Jobs Report, right? And I think the World Economic Forum also just published one last year. And in the latter report, it was said that there are, I think, 97 million jobs that were created, right, by AI and digital, 85 million jobs lost, right? Do the math, 12 million net improvement, right?
Simon Mueller:
But it's not that easy, right? Let's not fool ourselves. There's this massive kind of dynamic of transition, right? Anna, that you mentioned, which is a massive friction from getting workers where they are to where they need to be, right? It's not just doing the math and saying, "Oh, this all good, we're creating 12 million of net jobs." So I think there's a massive loss of skills, right? I think there's also just a lot of disruption and friction, right?
Simon Mueller:
And I think the other thing that's being said, when I see two tables, right? One table saying, "Here are the top 20 jobs that are being created by AI," right? And top 20 jobs that are being destroyed by AI or by automation." In the ones that are being destroyed, things rank high like factory workers, executive assistants, secretaries, etc. right? On the other table, on the jobs that are being created, it's things like Shireen mentioned, right? Advanced engineers and advanced technology, machine learning engineers, robotic engineers, etc.
Simon Mueller:
But I think the problem here is that those jobs are fundamentally different, there's a fundamental different quality of the job of an engineer, many of which are essentially creating machines that can create machines. So they're working on further automation. Same thing for data engineers, right? Or data scientists. There's this whole notion of we need more data scientists in this world, but I think one thing we're not really taking into account is the fact that... I don't assume that data engineers today or in five years are going to use pivot tables and spreadsheets, right? They will use tools that make it much easier to do data analysis at scale, and so where should all those 97 million jobs come from? So, I'm highly skeptical at that point, and we can talk about the gig economy in a second.
Chris Riback:
But what you were just saying, Simon, and the thought that we will be creating technology jobs that create more technology, as you think about opportunity, as you think about dislocation, as you think about disintermediation, as you think about the social drivers that are creating so many of the tensions in our society today, are you seeing us going towards a point where that opportunity gap is being closed, or are you looking towards a situation where things are getting worse before they're getting better?
Simon Mueller:
I think the latter, frankly. And the reason is that I think it's extremely difficult to upskill. So, A, there's this quantity issue that I've just alluded to, right? Which is, I don't think that there's just a quantity of jobs that are being created to make up for the job losses, I just don't believe those numbers. And B, the upskilling, I think is much more difficult than we imagine. It is not just taking a three months Code Academy or General Assembly Bootcamp.
Simon Mueller:
By the way, these are education systems that I like, because they are really innovative and to the point and very effective, but I just don't think that this can be done at scale, so I think it's a much harder and longer process than we imagine.
Chris Riback:
So in a moment I have a follow up on that for our labor economist, but before we get to Anna, Shireen, when you listen to what Simon is saying, I'm thinking about vertical farms in Compton. What's the training like? So you're talking about creating jobs in Compton, a place that is not terribly well known for farming. What are you needing to do in terms of upskilling in terms of training to be able to get those local employees who may not have background or learning or education in the areas that they are needed? Is there a training component that has to come along with building vertical farms in Compton, or elsewhere?
Shireen Santosham:
Yeah, of course. And interesting side note, Compton actually is historically an agriculture town. So historically, it was full of orchards, and the entire city, pretty much, is still zoned for farmland. There was a historical place called Richland farms that is still there. And so actually, a lot of people are very interested in farming in Compton, and parts of the city still have sort of miniature farms in them.
Shireen Santosham:
But the focus on training, certainly, and what we're providing are essentially high tech manufacturing jobs. So we're not trying to skill someone up to an engineering level, though in the long term it would be great to have a pipeline of people that could come into Plenty that have those skills from the local community. But at the moment, it's really jobs that are how to operate the robotics that are in our farm already. And so we have training programs that we'll invest in. We also are working with local workforce development agencies to find local talent and hire locally.
Shireen Santosham:
In the long term, this is a really good opportunity I think, for the private sector and policymakers to work together, because as you grow, especially for a business like ours, which is a physical business, it's not a software business, you need a pipeline of talent coming in. And so there's real incentives to invest in those pipeline training programs, whether it's with community colleges or otherwise. And you'll see that...
Shireen Santosham:
I mean, when I was in San Jose, we saw it with the airlines looking for airline maintenance techs, and so they would invest in training academies in the city, and those are $80,000 a year plus jobs. And so, I do think there's a really good opportunity here, it takes time, and there's certainly going to be friction, right? As jobs are transferred by location, right? So geography is certainly, you're not going to have exact matching on geographies, which is I think, the biggest challenge as we make this transition.
Chris Riback:
And Shireen, to that point, and when you talk about the partnerships required between private sector and public sector, what kind of tension do you see occurring or going to occur between municipalities, between the public sector outlets, maybe it's maybe it's San Jose versus Austin, maybe it's San Jose versus Compton, maybe it's San Jose versus Berkeley, is there potentially a shift? I mean, obviously, we know that municipalities have always fought over job creation, and they have always given various breaks and incentives to companies to come to their area. But given the gap that we're talking about, this pre-existing labor employment opportunity gap, given the importance of the types of jobs that you are discussing, these are next generation jobs, at least to somebody like me, what do you see happening between municipalities? And is there a way for government to get along when at some point there's a finite number of jobs that these towns are fighting over?
Shireen Santosham:
Yeah, well one thing I would say is if the Amazon campus sort of bake off contest, I remember a couple years ago that everybody might remember. Our mayor came out and said, "I'm not even going to bid on this, we're basically creating a situation where you're giving away so much value to a company that it's not actually going to be worthwhile for the city to participate."
Chris Riback:
But Shireen, isn't that easy for the mayor of San Jose, with so much going on in that area? It's a lot harder... Isn't that harder for a mayor who's not the mayor in Silicon Valley?
Shireen Santosham:
For sure, but part of that is by understanding the dynamics of these companies, right? So knowing what you're getting, because a lot of times what a company, Amazon or otherwise, will do is really create a very finite number of jobs and potentially bring in external folks versus hiring locally. And so there is some question about how much they're really contributing to the local economy. And of course, all these models, and Anna can tell us more about these economic model claims about how much value they're going to bring are sometimes pretty flawed.
Shireen Santosham:
But I do agree with you, I actually think local jurisdictions aren't, other than Silicon Valley, maybe New York, some of the big ones, are not well positioned from a tech policy perspective to really understand the impact on their cities, and really we should have more federal programs to help tap into, for any jurisdiction to really tap into, so that they can build workforce development programs that will service these sectors, right? Because ultimately it's about people and talent. They keep on trying to build Silicon Valleys in other cities, but ultimately, the valley still maintains, despite the incredible cost of living and challenges that come with living in this area, people still stay here because there's just such a large talent base that you can't find in other places.
Chris Riback:
So I agree with you. Let's turn to our labor economist and more, Anna. So Anna, first of all, please, any insights, analysis, maybe research that you have, I don't know if you do or not, on some of these dynamics that we're talking about and the pending, potentially, battles among municipalities. Personally, I wonder, Shireen's point about a federal policy to help guide, that would be terrific, because if not, I fear this potential kind of Lord of the Flies environment with all of these cities fighting over these valuable jobs.
Chris Riback:
So one, any thoughts on that? And two, maybe, what do you see in terms of, as these jobs get created, as these gaps get created, heightening calls for changes in the social structures, whether through a universal basic income or other redistribution. So two areas there, Anna, if you could tackle, plus, of course, anything else that you would like to add.
Anna Stansbury:
So a lots to tackle, Chris, but I'll try. So I think I fully agree with what Shireen said on the kind of possibly destructive nature of local municipalities competition for something like the Amazon HQ or similar things. And I think in a structure like the US where local business incentives and business conditions are so sub-nationally determined, you can get this vast power imbalance, where essentially a large company can capture all of the net social gain from its business by competing off different localities against each other. And so I think some kind of federal flaw or federal countervailing initiative to that, that would prevent type dynamic that you see, either with HQ competitions or with local business tax rates or other kinds of things is very, very important.
Anna Stansbury:
In terms of the kind of training and disruption conversation that we've been having. I think there are two things I just want to add to that, although I agree with what we've said so far. So I very much agree that training and retraining is super important.
Anna Stansbury:
One of the questions is who should bear the risk of dislocation? And our system right now in the US places almost entirely that risk on the individual. If you've chosen to enter and train for a career, that then gets automated or gets outsourced to a different country, essentially, the provision of any income support is very limited, the provision of retraining, there is some but it's very limited, and the decision as to what career path to go into can be quite difficult and complex, you're trying to take a 30 year horizon where experts don't even know what the 30 year horizon will look like.
Anna Stansbury:
And so, clearly government I think should be very responsible and should be providing much better income support for displaced workers and much better retraining and potentially job guarantee type services, including income support for people who maybe don't wish to move for communities that have been decimated, and enabling those communities to persist at least in some form.
Anna Stansbury:
But there's also a question of employment and employers and how much employers should bear that risk? There's an interesting academic study that I think we can kind of learn something from. So there are studies that look at the impact of robotization on factories and factory workers in the US and in Germany. And these robotization studies in the US are very well known and well cited, and the idea that we all know is, as robotization comes into manufacturing, manufacturing workers are laid off and communities that depend on manufacturing work often go into decline with persistent long-term unemployment or non-employment, rising debts of despair, declining local tax bases, declining local economic outcomes for the local service sector that depend on the spending from the manufacturing workers. And this is at the root of a lot of the social problems we see today.
Anna Stansbury:
In Germany, looking at the same types of robotization being introduced into manufacturing, you see quite different results, and in particular you don't see any increase in long term non-employment or unemployment as a result of this robotization. And one of the hypotheses as to why that looks plausible from the studies I've seen is that in Germany the factories where workers tasks automated still reduced their employment, but they did so entirely through not hiring new workers rather than firing existing workers, and we know that layoffs are one of the most disruptive aspects of technological change or other kinds of social change, because these are workers that are probably in the middle of their careers that have invested substantial time into skill development, and very hard to find a new job to move to a new place, whereas young workers are much more easy to train and retrain, much more geographically mobile.
Anna Stansbury:
And so thinking about what are the social structures that protect the workers that are already employed? These German manufacturing workers were reemployed in the firms doing different tasks, and how can we actually create an employment system policy wise, potentially that forces firms to internalize those externalities, those negative external effects that they create when they lay off large numbers of workers in kind of concentrated areas. So I think that's an important policy part of the conversation, and it feeds into another set of debates about high road versus low road employment strategies.
Anna Stansbury:
So in some ways it can be equally cost effective for firms to employ workers in a high low type employment strategy, where you have low turnover, you have high degree of investment in the skills and training of your workers, and the workers in turn know that they're not going to be fired, they're not going to lose their job easily and so have a high degree of investment in the firm and in the quality of their work. And then you have low rate employment strategies which essentially do the opposite.
Anna Stansbury:
And I think, I mean, it sounds very much like what Shireen is describing with Plenty as a high low type employment strategy where you're investing in people, investing in their skills and in turn reaping the benefits of people remaining at the company for a long time, promoting up through the ranks in the company and creating good pathways to good jobs. And so I think thinking about how employers can choose to, but also how policy can encourage employers to pursue strategies where they invest in their workers and have a long-term sense of obligation to their workers is important.
Anna Stansbury:
The final thing I want to say on the skills point is, a bit of a challenge to this framing of the idea of the skills gap and the importance of skills and retraining, which is we have millions and millions of jobs in this country that are not going to be automated anytime soon, that require in my view, quite important and substantial skills, but which are often discussed as low skilled. Thinking about six, seven million home health and personal care aides, childcare workers and nursing assistants, thinking about 3 million janitors, maids and housekeeping cleaners, thinking about a lot of food service workers, garbage collectors, construction laborers. Many of the lowest paid jobs in our workforce are not going to be automated and will still need to be done.
Anna Stansbury:
And so in some circles I think the idea of re-skilling and up-skilling can be seen as a bit of a panacea to the problem of income inequality, but if we're going to have tens of millions of jobs that need to be done, regardless of whether someone has a coding degree or not, what we need to be doing is thinking about how much do we want those jobs to be paid? How good do we want those jobs to be? And how good we want them to be in terms of working conditions and how much do we want them to be socially valued?
Anna Stansbury:
And so that brings to the last point, which is, I think the question of redistribution is fundamental, I think the question of pre-distribution is also fundamental. So not just redistribution next post with taxes and transfers, something like more progressive Taxes, UBI, but also how do we make these jobs pay well in the first place? And so I see a very important role for unions, for minimum wages and for other ways to empower the workers to kind of in particularly low paid, poorly valued jobs in our society to have the power to organize for their own working conditions and pay.
Chris Riback:
Well paying jobs is something that many of us could support, isn't it? Two notes, one, for folks listening, surely a bunch of ideas going through your head, questions, please get prepared to ask them. In a moment we will go to audience questions, please note the webinar is being recorded and will be posted later online. And to ask a question, we ask the participants, use their raise hand function. Rachel will notify you when it's your turn to speak, you may experience a short lag time, we haven't fully solved every AI machine learning new technology problem on the face of the earth. Simon will be right on that and have that solved by the end of the conversation.
Chris Riback:
So if there is a little time lag, please be sure to take yourself off mute and turn your video on when prompted, also please state your name, degree and year of graduation before you ask your questions, and please remember all questions really should end with a question mark. Before we get to that part of the program, Amy you did many services for us, one additional service that you did was you helped segue to the next thing I wanted to ask about, which was, you talked about places outside of the United States, and I'm curious for each of you, somewhat quickly, because I do want to make sure that we get questions from the audience, should any exist.
Chris Riback:
Are there opportunities for emerging economies to take advantage of the opportunity gap of doing up-skilling training on their own and coming in and being able to compete and win jobs? I.e. To what extent is there global opportunity or potentially this is the contrary, are companies in the West still a bit averse to distributed work? And perhaps equally importantly, are Western governments becoming more averse to job creation where the jobs don't sit within a country's borders? Anna, why don't we just ask you very quickly on that and then, because I wanted to get to Shireen and Simon on these issues as well.
Anna Stansbury:
Yeah, such a great point. I think this is a hugely exciting opportunity from the perspective of global income growth and poverty alleviation. The fact that distributed work will enable workers increasingly from all over the world to do higher value tasks in the global value chain in a way that we've already seen to some extent with offshoring of business process support and IT, that we're going to see that with many other tasks. You can think about, medical services, not just the interpretation of lab results or radiology scans, but also potentially the provision of video consultations or chat consultations. You can think about education, you can think about many of these.
Anna Stansbury:
And so I think the potential for the net expansion of human welfare as a result of this is huge, and yet, again, the question is going to be, how do emerging economies take advantage of that best without a disruptive competition between them and the way that we've talked about the destructive competition between US localities for a scarce set of good opportunities, and how do we deal with the ensuing disruption of what were formerly well paid jobs in higher income countries.
Chris Riback:
Thank you. Shireen, thoughts?
Shireen Santosham:
Yes, absolutely. I mean, as related to this conversation on AI specifically, I mean, one of the things that is already happening is the outsourcing of sort of fundamental sort of database inputs for AI models. So for example, for image recognition, you need to have these massive databases of like pictures of dogs, whatever it may be. And those need to be very laboriously coded by hand manually before you can actually do any analytics on that data. And so you have these big facilities across India and different parts of Asia that are doing this already, so you are seeing that kind of work happening and it's important to understand that the technology, although it can display some jobs is creating other kinds of jobs that is an important source.
Shireen Santosham:
And those jobs sometimes are problematic if they're disturbing images or all sorts of things, side effects that we're learning about. And so I do think that's an issue. I think the biggest, from my perspective the base issue frankly is just the amount of money that China specifically is investing in the technology sector, whether it's agtech or otherwise, it just is outpacing investment by the US and other countries by orders of magnitude. And so just as we've seen kind of these huge large tech companies grow up, and now we're having a conversation around whether they have too much power in the United States, think about that on a sort of political sphere and what is going to happen once China sort of gets far enough ahead.
Shireen Santosham:
Even today about a third of the workforce in Silicon Valley in tech companies is Chinese or Indian, so they have the talent and we're even using the talent for our companies here. And so, I think that's the biggest question on how we sort of assume we're going to be the leader in AI, but will that shake out differently over time and will we actually see a different balance of power?
Chris Riback:
Well, and especially given Xi's proclamations of, I think it's the 2025 plan, if I'm not mistaken, ensuring... Simon, well, correct me if I have the plan wrong, but Xi's statements around the role that China is going to have, wants to have in terms of the new technologies and new opportunities. Simon, thoughts that you have on the role that emerging economies or outside of Western economies might have in terms of perhaps taking advantage of these gaps and these new opportunities?
Simon Mueller:
Maybe just anecdotes Chris, for the Future Society, for example, we are a completely international and global and remote organization, and we've been doing that for the last five years, right? And I think now the world has woken up to the possibility to work remotely from every corner of this world, so I'm just ramping up another venture right now, and our designer is in the Ukraine, right? And so the possibilities for good talent that is outside of the country you're doing business in right now is just fantastic to be able to work remotely and contribute using the modern technologies, right? It's kind of outside of our AI necessarily but I think the ability for talent to be truly mobile, I think has definitely increased with the technologies.
Chris Riback:
Thank you. We are starting to get questions from the audience. Romi Amin has a question. We will give Rachel a moment to promote Romi, and by promote I mean, get Romi a new job, higher pay, more benefits, all that kind of thing. Romi, are you ... let me see here.
Romi Amin:
Hi, Chris, can you hear me?
Chris Riback:
We can. Super, thank you.
Romi Amin:
Fantastic.
Chris Riback:
If you would Romi, your degree and your year and your question.
Romi Amin:
Sure. I'm happily working at the world bank, no job requests. I graduated from Kennedy School about 10 years ago, worked a bit at Shorenstein center and Belfer center and now working at world bank on many of these issues. In fact we've worked with the Future Society, Simon, with some of your colleagues. As much as I have so many questions for you, I might focus questions on Shireen and Anna.
Romi Amin:
Much of our clients at World Bank are low and middle-income countries, and I'm grateful Chris to you for that last question of trying to get into some other countries beyond the US, although the discussion when it is sort of the emerging and the sort of Indias and Chinas, and one of the questions that I have maybe for Anna, from the ongoing discourse within broader labour economics, labour market displacement, what's the research looking like for low and middle-income countries, what conclusion are you reaching with that sort of research? How robust is that research? I'm finding a lot of the studies I read are outside of that context.
Romi Amin:
And then also Shireen, if you could also chime in on this discussion, I think your venture sounds fascinating, this AI and Agtech, but I'm also curious about within low and middle income countries, if you've considered how would you speak to a government who's hyper concerned about having massive employment in their country with agricultural labor, that is very concerned about what a couple of AI factories might do for this huge sector in their countries. So I'd be curious of what you might advise to say to a government that's concerned about these sorts of things within our kind of world bank framework. Thanks very much.
Chris Riback:
Thank you, Romi. Yes, Anna, please, why don't you start and then we'll go to Shireen.
Anna Stansbury:
Happy to. Thanks for the question. Yeah, so my sense from the research, this is not a full overview is there's kind of one big threat and one big opportunity for emerging economies viewed from the 30,000 foot level. And the big opportunity is the extent to which AI will expand the ability for distributed work to be done and reduce things like language barriers and time zone barriers that inhibit distributed work from functioning well even now, and so that expands massively the possibility for essentially export oriented jobs in the sense of service exports that low and middle income countries can then structure work forces around and structure development strategies around.
Anna Stansbury:
And I said something about that in response to Chris' question already. So I'll say a little more about the threat. The threat I think is the question of whether the automation that is being done of tasks is actually appropriate for the world as a whole? And the question here is really who is doing the automation? Who's designing what tasks should be automated in the labor market, and what incentives are they facing.
Anna Stansbury:
So, economist Lant Pritchett has a phrase that always sticks with me which is, why are the parking ticket kiosks automated in Uganda? And the idea there is that you've got a bunch of research scientists in the US trying to figure out how to automate tasks that are relatively expensive in global terms, because they're being done by workers who are paid relatively well in global terms even if they're paid poorly in the US. Self checkout machines, parking ticket kiosks.
Anna Stansbury:
But in a poor country, those jobs could actually be better paid valuable jobs, but the technology once it's developed is then quite cheap to implement even in poor country contexts. And so sort of getting rid of jobs before they've had a chance to use those jobs on a stepping ladder for economic development.
Anna Stansbury:
There's this whole conversation about premature de-industrialization, the idea that the automation of manufacturing means that for many countries, export oriented manufacturing led growth strategies. are not going to be strategies that generate mass middle-class employment opportunities anymore, because manufacturing is so much less employment intensive than it used to be when it was the development strategy, not only for the industrialized West, but also more recently for sort of the East Asian tiger economies.
Anna Stansbury:
And so worrying about that with the next wave of automation of middle skill, good service jobs, and whether there'll be a kind of premature de-industrialization equivalent in services, because the automation is being done to respond to labor cost conditions in rich countries and not in poor countries. that's what I worry about the most.
Chris Riback:
Thank you. Shireen, any thoughts on that or perhaps comments to the fictional government that Romi suggested you might have to have a conversation with?
Shireen Santosham:
Sure, and thanks Anna for mentioning Lant, he was one of my professors when I was a student, so I hear him when you made that quote. In terms of our goals at Plenty and how to think about a technology like vertical farming. I would say one is, it's not a panacea, so it's not meant for every space and every situation. And so as Anna was saying, it sort of like what is fit for purpose for your country or economy? I mean there are places where things like vertical farming are going to be especially important or places that are water scarce or food insecure, or have the need they're already seeing disruption in the amount of farmland that they have or reductions there. Certainly the water crisis is something we don't talk about enough, we're going to see massive water shortages around the world in the next 20 to 30 years, including in the United States where we're running out of groundwater in some of the biggest farmland areas in the country.
Shireen Santosham:
And so I think where we see a lot of demand for this kind of farm is in a place like the middle East parts of China, you can think about Island nations oftentimes approach us and ask us if we're willing to go to them because they don't grow fresh fruits and vegetables and you know how much it can cost for them to import.
Shireen Santosham:
And so, I would say this should be part of a larger agriculture strategy for any country and as you're really thinking about what your population is going to need over the next coming decades, it certainly should be a consideration in how you think about food security.
Chris Riback:
Thank you. To your point, Shireen, I think there was a piece just yesterday, a Bloomberg piece that said, I think it's a 20% of ground Wells are empty or nearing empty or something. I think it was in the US I'd have to look at them, I'm close on the STAT, but I'm pretty sure it was a Bloomberg piece just yesterday to your point of the running out of water. So I'd like to try to squeeze in two more questions, both very important topics, one is an issue around AI that we haven't talked about yet, but is very much in the news and that's bias.
Chris Riback:
And I'm wondering to what extent you all believe and obviously this goes directly to labour and employment and opportunity. To what extent do you believe that AI will remove unconscious bias in the workplace or are the algorithms themselves designed with bias? Obviously we all have been following so many of these issues around AI and bias the events at Google have been most prominent but it happens elsewhere as well. Shireen maybe we could start with you thoughts about AI design and bias.
Shireen Santosham:
Yeah, I mean, I think right now we're in a situation where there is unfortunately a lot of bias in the AI, and it's not necessarily intentional, but we've seen pretty high profile cases like the Amazon facial recognition software that was run against ACLE ran it against Congress and discovered that it misidentified 28 members of Congress as felons, including John Lewis, the famous Civil Rights leader and of course it was more biased against people of color.
Shireen Santosham:
And so these types of tools are able to be purchased by police departments for less than $50 a month with little to no oversight, and there isn't a lot of accountability, frankly, at these big tech companies around this issue, so I am highly concerned about this, I think there's really good academic research happening to look at this, but places like Google, where they have an ethical AI group we've seen the high profile firings of Timnit Gebru and Margaret Mitchell.
Shireen Santosham:
And the question of, yeah, you can publish a bunch of principles, but how does those principles transfer into your product? To be fair to the companies, it's actually really hard to do, because sometimes you have unintended consequences. And so how do you create frameworks that test and give you feedback and redesign products without sort of these large impacts? Because the problem with tech of course is that it accelerates and amplifies negative externalities and it happens really fast. And so I don't have a solution there, but I would say we are in a situation that is highly problematic, it's probably going to get worse before it gets better, and I hope we can find a way to govern this better.
Chris Riback:
Yeah. The things move very, very quickly. Simon, Anna, either of you have something that you want to add on that point, otherwise I have a question about capitalism to try to close out our conversation today. But anything on AI and bias that either of you want to contribute?
Simon Mueller:
Maybe just, the key issue of bias is that are encoded in algorithms is really the training set, right? It's the training data. And so if there are stereotyping involved in the training set, right? You get prediction machines, rather prediction algorithms that essentially are as bias. So I think there's a lot of work to be done to open the black box, to understand the structure of the data coming in to be trained.
Simon Mueller:
But I think the other important thing is we shouldn't fool ourselves, right? AI is not just a tool, it's a social technology. So it's also people using algorithms, using AI, using them as tools and that is the problem and I think we get more and more dependent on algorithms and we are less and less kind of able to really question inputs and outputs and inner workings. And so I think we have to really understand, I think the awareness of what algorithms, what the structure of algorithms are and the idea of it and the mechanics behind it, I think it's tremendously important. Otherwise we'll just follow all types of suggestion and prediction engines as we do on Google maps, right? No one is doubting a route that's been calculated by Google maps, because it happens to be always right, but we have to really be aware and critical.
Chris Riback:
Excellent. And if we could close out maybe just 30 seconds from each of you, really around what's next, and what's foremost on your mind, what are you looking at? A question came in, it might frame what you're going to say around do you agree that any symptomatic problem arising from the optimization of AI in the economy is mostly inherited by a problematic root cause found in the capitalist frame it operates within? So I think I understand the context of that question, but if you want to address that one or more broadly, what's next? What are you focused on AI? We've got kind of 30 to 45 seconds each, Simon maybe you can kick us off on this final round.
Simon Mueller:
Yeah, one thing I'm thinking of as we think about how capitalist incentives are shaping how AI is being used, right? I mean, I don't even know where to start, but one thing I'm currently thinking of is the impact of the gig economy and widening gig economy, where you have a central kind of cognitive overload that is allocating work, in fact tasks, right? As efficiently as possible, but that system, right? Has all types of positive effects, right? Such as Uber driver can choose when to work and for how long to work, et cetera, is all great, right?
Simon Mueller:
But I think the power imbalance that I think Anna is mentioning so often, I think it's just being exacerbated by the fact that cognitive function and both sensoric and actuating functions, like a human actually doing things like driving the car, picking the groceries, et cetera, is further dis-intermediated, just creates a tremendous amount of problems and I think hiring or contracting them as kind of independent contractors is probably the wrong way to solve problems.
Chris Riback:
Thank you. Shireen, closing thoughts.
Shireen Santosham:
I think the thing I worry most about especially over the last year has been sort of the rise of the conspiracy theories and the impact of AI and sort of bringing people down this road of more extreme thinking. And this can feel abstract in some ways, but when I was working in city hall, I mean, people show up, right? They'll show up and they'll be really concerned about a 5G cell that you're putting up somewhere because they believe it's going to give you cancer or something else and they're really dug in on some of these issues and that goes all the way up to [inaudible 00:57:29] and these other conspiracies that we see.
Shireen Santosham:
And that combination plus growing inequality I think is what is really potentially what blindsides us, because we're all in sort of our own little echo chamber of tech without sort of crossing those boundaries.
Chris Riback:
Thank you, Anna, take us home, please.
Anna Stansbury:
Thanks. Yeah, so I think the biggest thing I want to sort of caution against is a sense of technological determinism or fatalism that the path of technology is going to shape the labour market in a certain way, and we can't respond to it. And I think this conversation and the framing, Chris, that you've brought has been very much against that view and I'm really appreciative of that.
Anna Stansbury:
The other point is that I think we need to be ware of bad automation or sourcer automation or not innovative technology. And I think there are two good examples of this, one is the example that Simon talked about the gig economy. We have a really great new technology that can match tasks to be done with workers to do them, but the biggest innovation of the gig economy in my view has actually been a really successful strategy of avoiding labour market regulation, such that now taxi drivers, delivery drivers don't need to receive the minimum wage, health benefits, workers' compensation, unemployment insurance, and all the other things that make employing people costly.
Anna Stansbury:
And so making sure that the innovation that we are pursuing and rewarding in society is truly pie expanding, and then we figure out how to distribute the gains, rather than innovation that's just seeking to disrupt protections that we've put in place for good reasons is I think a sort of salutary note to leave on.
Chris Riback:
Man, the three of you just set up three new Zooms on AI on the topics that you raised. Anna, Shireen, Simon, thank you so much. Karen or Rachel, any closing words to goodnight everyone?
Karen Bonadio:
Yes. Thank you, Chris, and thank you to the panelists and to all the alumni who joined today, we hope you have stayed engaged through the alumni top policy discussions and look forward to your participation in other opportunities. Please join us next month for the state of the school address with Dean Elmendorf for the inaugural meeting of the alumni association, hosted by the HKS alumni board. For the most up-to-date school news and events, please visit our website, stay healthy and safe everyone. Thank you again.