Transcripts For CSPAN2 Mary Gray Ghost Work 20240713 : vimar

CSPAN2 Mary Gray Ghost Work July 13, 2024

Faculty member of the Economics Department and the cochair of the mit work of the future attack force that was commissioned a year ago the spring and the purpose of the task force is to engage in confront the sense that many have that the labor market and technology is changing rapidly to understand why people are anxious, to figure out how concerned they should be, what is changing and what is different this time and what can we do to ensure that the coming era of change needs to not productivity growth but some that show up in our focus is on the labor market and how we make the labor market function well for as many workers as possible. Thats where we see the opportunity and the challenge in the lesson if you read the Task Force Report which was released last week, one thing that we s say, i think a critical economic lesson of the 21st century, a wellfunctioning labor market is the foundation for a healthy middleclass in a wellfunctioning political system. Without that foundation of opportunity and prosperity is difficult for many other things to work well. Our focus is working on making work work and thats why its an honor to have mary gray here from the university of research, talking about her book ghost work. Heres a photograph and heres the actual person and just to give you a tiny bit of background, mary is a cultural anthropologist whose work has been on the communication and use of technology of young queer people and how they use technology and other means to establish identities, to identify one another. She was recruited and once there she got interested in the question of how Artificial Intelligence was said by workers, often times hidden inside the machine. So you can feel the uniformity and this is kind of the intermediation between machines and people and often identities that are completely obvious to the participants involved. I read her book and i thought it was quite ongoing and some of the things i have been doing and we are delighted to have barry here, shes going to speak to you about 25 minutes about the book and then there will be questions answered future pencils ready and will be delighted to hear your questions. Thank you for indulging me and giving me a lecture to set up the conversation. Im just going to stop for a Second Period how many of your part about the recent decision in california. I hope that comes up in q a, primetime. How many have you have ever received the 1099 or done contract work . Wonderful. , how many of you lets make it easier, how many do you you have a friend that use of ridesharing app called uber or lyft. Perfect. With that, we all have a stake in this conversation. Arguably. Or definition, Michael Arthur came up with this turn not to describe particular kinds of work, not about a particular set of mixed jobs, arguably to describe the reorganization of fulltime employment. What we want to think about where the conditions that are created through the opacity around workers contribution in the value of ondemand work so keep in mind, hopefully you will notice in the book, they you will never be a reference to go to the workers, youll see references the ghost work and that was intentional and of course like anything after you publish it you dont control after words out of your hand. I want to give you an idea of what the mechanism behind the work we are talking about and what connects this work that i will describe two apps and other kinds of platforms that you might be familiar with. What you want to understand, the early 2000 computer, scientists and engineers realize they can take the same sort of mechanism for calling and putting out a particular call interfaces, apis that would draw up software to execute a program and they can spread people into the executable. So they quickly ended up creating a mechanism that could help them with their work which was offered around 800 data labeling and other kinds of tasks that was piecework, youll be familiar with amazon more than any but we will suggest this is just the beginning, this is the piece that you can see. But as any business that has a task that wants to demand or request to help with. Something that can be put on a platform between the relationship between a twosided market, and an abstract thats works going on with the mechanism. This is any taskbased project driven work that can at least in part be sourced, scheduled, managed, shipped and billed through an application programming interface to the internet. Hold onto that and think about how much world of work it could develop if youre applying the mechanism to the distribution of request for work and people being able to pick up the tasks. Thats what i want to suggest to you is the framing we should use in the future about work, not automation and if anything, even laster talking about the future of work was often like robots, we started thinking differently about that but in many ways we have not come to the floor with the framework if its not robust, what is it, were suggesting is potentially ghost work and we need to stop it before its built into everything that we do. So importantly you are probably familiar with the category of online offline Platform Services anything from uber, lift, doordash, then using the same mechanism, and api application interface putting out a request for somebody to pick up the food, deliver it to this address and the platform is participating in the exchange by recording when they picked up and delivered, executing a payment, scheduling where it will arrive, giving an address, that portion of the work is automated but the delivery and the value of somebody being there to deliver the food, thats the equation we are not considering or more increasingly were aware because we can see the folks, if i had said content moderation was a form of the labor i dont think anybody wouldve known what i was talking about until analytic type. You will know the content moderation of the job that people do that is absolutely providing another service which is training Artificial Intelligence but importantly those are people on an executable loop performing a very important service. We are focused on the vast World Business start up, often businesstobusiness services that will be below the surface, if anything you oversee as a consumer. That is the world of work im going to talk about today. Is the world of editing, the world of testing, data labeling, perhaps some of you in the room, many of these different tasks do drive Artificial Intelligence innovation, instructions a clean data set and i love being in it because most in the room know what im saying when i say that. But importantly in increasingly seen the number of jobs, it is quite hard to nail the a. I. But will just keep a person threaded into a Service Request which perhaps textbased, anytime you got onto a website and have a little help window pop up, you know its a mixture of script and a person who is assisting you. So in thinking about that world of work, when you have a person doing something on the spot that cannot be quite completed by automation, theres probably points of reference that come to mind and we cover this in the book, places to thread together, this is not new, the continuity and how we treated people who would be a moment because automation would come around and they will be replaced. That might look like piecework when in the days of textiles certainly those who have had manufacturing knockout are short but couldve added the button or the bow, for quite some time, usually that was a matter of decades, thats the important thing to take away, automation made it possible for piecework to go away and textiles in some cases and other cases both labor cost but also the reality of the paradox of automation last miles being too sophisticated for a text file. Machinery to consume their automation. Then a person was kept around. Its also in the world of federal Contract Labor. The example of the women who were made famous in the films in the figures who could at the time be brought into service, computers which we again at mit know that was a reference to the people, not the machine and that was quickly when the need in the demand for their services as computational experts eventually disappeared, they could be let go. There was a security and that employment, that it was not less valuable, we didnt have a way of looking at unemployment and seeing the women as valuable pricinprecisely because we move forward of what it was to value working a fulltime employment in a particular profession and often particular bodies embodied in those professions, white men of privilege who had very specific roles to play, anyone who not playing those roles seem dispensable. To continue the lineage by the 1960s and the advancement of staffing, that quite literally, i point you to the beautiful book on the economy, quite literally brokered on the value or the devaluing of womens labor as a resource because they go after young woman who is collegeeducated and make great office girls, a nonoffensive term of the time, there also expendable. So keeping the thread moving, by the time we get to the 1980s and two thousands in the outsourcing and off shoring of knowledge work, it becomes much harder to make the case that people are doing something that could so easily be replaced, precisely because theyre doing work that is being done by workers in the United States, it becomes much more obvious this is a question of labor offer trash, you can find cheap labor that is just as educated as anyone that is generating the request for work and i often met some of my colleagues, the settlement of the case against microsoft that involve microsoft, meant that we never resolve the question of what do you do in the case of employment that is necessary for a period of time, project driven, you might need somebody with a specific expertise, language expertise, but you know you are not going to need them for more than 12 months or 12 weeks or 12 days. What ways do we have to value that worker. And at the time we really did not have a category four that. Its important to note the post 2000 Silicon Valley and especially 2001 what happens effectively the. Com bubble burst then at that moment we resolved and settled without case law that were going to leave questionable what to do with people as necessary but not necessary as were going to hold them for career and what we came up with in the settlement were a set of practices that treat Contract Labor through vendor Management Systems that often dont leave them with the protection beyond the 12 month contract to be able to say unemployed and these benefits come with my employment. If you think about the history that weve drawn and it certainly an argument, any historical trajectory is an argument, in this case we see the setting in place from the beginning of the industrial era of laws around labor protection that mostly assume the valuable work is the work that cannot be automated without much projection that may become the target of automation. So policies built for Assembly Lines and professions that were beyond the touch of automation and in the whys of Temp Staffing which driven our Economic Activity globally is the growth of a Service Industry that serves people request for need more than a serves the need to build something. And lastly to see the shift towards arguably the information economy that involved people doing Information Service work. Yes it involved people doing coding and other valuable skills to take great amount of training but if you think about what it took to code up a website back in the 2000 and you have the hand coat html, how many people in the room know what im talking about, that is done with software. And we paid quite a bit of money to build peoples website, for my first work as a 1099. Keep that in mind as you thinking about what can be automated, what is in the work of the creative look, the complex communication that seems beyond automation. Thats an open question. What is it that will be constantly on the horizon that requires human touch for some amount of time although we cannot see growing into a career. With that, let me give you how we studied this world, studying is a distributed world of work, not the most obvious thing to do, it was hard to figure out where to begin, we chose for businesses as case study to be where we identify groups of workers and also these companies to show the inside of the black box, how did they organize the work, what does the workflow look like and for the world of the mechanism that can build out Artificial Intelligence but keep humans in the loop. Theres two things of work, in terms of framing if there is something i hope you take away to see that the growing world of work is really growing into different directions, one is increasing need to structure and analyze data and take the example of a radiologist, using a. I. And health and radiology is not a one and done matter, you might be able to do a great job of getting Image Recognition for particular problem with him radiology to seemed like theyre going to get rid of the radiologist. You just expand the market for people using radiology especially to create mechanisms for getting outside of urban centers that is the places where no one has been able to get access to a stream and you also created a market for other kinds of diagnostics, you gonna require amount of domain expertise by medical professional. It is not a doctor doing that work or radiologist, and might be a medical student or a new profession on the horizon. So thinking about that world of work that is managing the information and data that we are collecting, that will build Artificial Intelligence and take over the need for human hand in the mix, but then the second stream which is just as critical, human in the loop Information Services what we come to expect as consumers but someone who will be able to answer my call 24 7. I would love it if none of us had the expectation. I would really love to feel that somebody is not inspecting of me. The reality of a number of Small Businesses that want to offer you a prompt to take your medication if youre taking care of an elderly parent and what that prompt be sent to the parent but also something to tell you how their care is going today, those cases are the world of human in the loop that are not fulltime employment, the pastoring, project driven work. In the case of our study, the two things of work are analyzed through the cases of companies that both do image tagging and verification, we pick these things so people thought it would be easy to automate, the not so easy to automate. And then contact moderation is another example of the stream of work that we hopefully would be able to remove human hand with automation and content moderation is a fantastic boundary object that says its really hard to get rid of people from the task of the viewing content. Because anything that is not an obvious black or white is something that is very difficult to automate. You have to have yes no, you have to have four things that tells you this not that, if you have humans deliberating the hate speech or love signal. And we debate that. How do you classify when is that hate speech and when is not a moment of intimate loving exchange. Content moderators are a part of the solution for being able to facilitate and identify what is going on in that moment. Also social media, we dont have to worry about this. There is not option. And then the second set translation and captioning a video in a generation, these are wonderful cases like this is a business, businesses serving other businesses. Theres a lot of Economic Activity happening. So we chose the United States and india because we wanted a complicated project to work on and keep doing work for seven years. And no notice some patterns in much of the cases with india that we met in the south and the conversation about that, theres less obvious of a pattern in the United States im looking for anybody who would like to keep looking for the analysis of the data sets because we barely scratched the surface of what patterns we can see but importantly the material comes from thousands of interviews im sorry thousands of surveys and hundreds of interviews. Were in peoples lives for that pattern and their routine in and out of these markets. In doing the skilled work, it was also an effort to figure how do we look at the largescale data that these companies are producing. We are able to get to the transactional data so we can see how many clients are requesting the services, how long do they stick around, do they change what request they have . I want to leave you with two findings, hopefully this gives you a taste of whats in the book. For me the most pressing the first like any open call, any environment we are saying there is no obligation, and fact there is no commitment but if you like to come in and do something, come on in. Guess what happens, you end up with folks who are really in it who want to make it a fulltime income stream that theyve identified as enough money to make the rest of the economic reality manageable. Because you have a percentage of people that will see value in saying and will turn to this work. Then you have another good core 20 who are saying im going to do this in a set amount of time, they have their reasons and ill get to that in a moment. They are the deep backup, the bench that is there and when the 10 is walk away because her kids are sick and they need to do something else, you as a consumer will never know that the 10 is really good inefficient as executing on a Task Force Project ever walked out. Because the regulars are there effectively making themselves available to be ondemand to being consumer. Lastly most importantly, this is a lesson that feels really important to me in the wake of the california case, you have a long tail of people were experimenting with the labor markets, they were particular kind of capacity or capability that theyre bringing to the table maybe they want to trade debugging or coding, maybe they want to practice the language and they want to see if this is more interesti

© 2025 Vimarsana