The Future of Work: Upskilling Deskless Workers

UC Berkeley Extension
32 min readMay 31, 2024

--

Welcome to The Future of Work podcast with Berkeley Extension and the EDGE in Tech Initiative at the University of California, focused on expanding diversity and gender equity in tech. EDGE in Tech is part of CITRIS the Center for IT Research in the Interest of Society and the Banatao Institute. UC Berkeley Extension is the continuing education arm of the University of California at Berkeley.

This episode, we’re turning our attention to the deskless worker, those who work in manufacturing, frontline support, health care and so many other areas. Their job does not require them to be at a desk or in a traditional office setting. They’re on the go, in the field, working in environments that demand mobility and flexibility. How do we accelerate the learning and upskilling of the deskless worker? Is it through AI, advanced communications, introduction to new opportunities? How do we incorporate more inclusion for this majority group of workers?

To discuss this important topic, we’re delighted to speak with Muriel. Muriel Clauson is the co-founder of Anthill, which is an IT governance platform that fuels AI adoption for non-technical workers. Anthill is focused on helping deskless and overlooked workers have the same support and access to technology in their careers as their desked peers.

Muriel has been an active researcher and an advisor on the future of work and artificial intelligence. She has authored scientific publications, was a scientific advisor for several governments and organizations around the world, and is an Advisory Board Member for Humans for AI. Welcome, Muriel.

Muriel Clauson: Thank you so much for having me.

Jill Finlayson: Well, this is an exciting topic, and I was hoping you could help set the stage for our conversation by defining the deskless worker and how many folks fall into this category.

Muriel Clauson: Yes. And so first of all, why are we using this word “deskless?” So there’s a few synonyms you can think of. Frontline, historically referred to as blue collar. You can think of these workers as folks who, as you said, don’t sit at a desk or computer to do their jobs.

And when we actually asked a whole bunch of these types of workers, hey, what word best describes what you do? They agreed, deskless was the term because they said they were free of desks. We’re out there in the world. And I love that idea of the empowerment that comes with that. So we call them deskless workers. That’s the word that I like to use.

Deskless workers represent over 80% of the world’s workforce. So this is the main way that a human being on planet Earth works. And I think it’s really easy to overlook that. A lot of the people who do the research, build the software, come up with the HR programs in our organizations aren’t deskless workers and maybe haven’t been since maybe their first job.

And so it’s often, I think, an overlooked type of work. It’s a type of work that maybe we’re undervaluing in our organization, but this is work that keeps our world moving. So deskless workers sit across eight industries. Over 90% of employers employ at least one deskless worker. Even countries that we don’t think of as mostly deskless work countries typically have a majority deskless work population.

So the US, Brazil, many parts of Europe all have, at a minimum, 60% of their workforce working in a deskless capacity. So these are folks that are doing important work. They’re everywhere. This is the main way of working and we don’t talk about it enough.

Jill Finlayson: I like the freedom that comes with what you said. That they’re not tied to a desk. They actually have freedom and flexibility. People think about restaurants and hospitality, but there’s construction. There’s people who are working in manufacturing. Shipping. Drivers. Say a little bit more about all the variety because this isn’t one type of worker either.

Muriel Clauson: It’s a big group. It encompasses a lot of types of work. But the main distinction is between this idea of desk work, meaning I sit at a desk, I use a computer to do my job. I interface and communication technologies for most of my day, which is important, and we can get into why that is later.

And then deskless work is really everybody else, folks that aren’t necessarily sitting at a desk or at a computer to do their work. So you’re right, they exist in many types of jobs and many paths to get into those jobs. I love to use the example of the pandemic to think about the reality of this work.

So one of my favorite things is that a lot of conferences right after the pandemic, the speakers would be like, man, we all worked from home the last few years. And that’s just not true. That’s not true. Less than 20% of people worldwide ever worked from home during the pandemic or ever will because there’s many, many jobs where that’s just not possible.

And so think about those folks that weren’t actually working from home during the pandemic, folks that were preparing your food in agriculture, folks that were producing goods and manufacturing plants, folks that were moving goods around and supply chain, folks that were servicing people in all kinds of service roles. Hospitality. Health care.

There are so many important functions that folks are doing that we all really rely upon, and we see the impacts really quickly in our world when these kind of jobs are understaffed.

Jill Finlayson: And what are we seeing? Are the number of these jobs increasing, staying the same? What do you expect to see happen to this 80% of the workers?

Muriel Clauson: Well, so there’s an interesting phenomenon going on across the workforce right now globally. Just on the whole, the world is getting older. We have more workers over 55 than we’ve ever had. We have fewer workers entering the workforce globally. And so in general, that’s having a really interesting impact on particularly deskless work.

Right now in the United states, there’s about half a person for every deskless job an employer would like to fill. It’s about 0.52 is the statistic. And so that’s having a really interesting impact on all of us. But it’s interesting, I think, also to look at, well, why is this happening?

Well, one reason is that older workers are overrepresented in deskless jobs. So there is a pretty low replacement rate for younger workers entering the workforce. And if we look at the best projections from Bureau of Labor Statistics and a few other bodies, if we look at 2030, which is pretty soon, and we even factor in to recessionary events, during that time, we’re projected to add about 12 million new jobs in the US. If you look at folks who will retire, folks who will enter the workforce, we end up with about 1.5 million people.

So we have 1.5 million people for these 12 million new jobs. So jobs are growing in general. But then on top of that, the replacement rate for deskless workers is projected to be 20%.

So if you crunch all those numbers, basically the punchline is, imagine today we have one deskless worker for every two jobs, we’re essentially going to have one deskless worker for every eight jobs an employer would like to fill, which I think is very exciting because this is an undervalued type of work that we are going to have to start to value.

There is going to be an economic imperative for us to compensate well for these jobs, to treat people well in these jobs, and I think that that is a very good thing ultimately for people, because I think these are jobs that can be good jobs if we start to value them more.

Jill Finlayson: What do you think are the most common myths that people believe about deskless workers that we need to set them straight on?

Muriel Clauson: I think that they’re bad jobs. So any job is a bad job if you’re not compensated well and you’re not treated with respect and given the support that you need to succeed. Any job is bad. I don’t care if you’re making a bunch of money, even if you just have the other two, I’m not giving the support I need to succeed, I’m not treated with respect, that’s going to be a bad job.

So in general, folks assume that deskless jobs are not good, and often because of the compensation side of things. But there’s actually a lot of interesting research that’s shown that some of the boundaries that you get to have mentally with work are so much healthier in deskless jobs. So think about if you have a desk job and you’re working on projects or teams or collaborating, you’re reachable by email all the time, you are never not working. You are never emotionally fully removed from that job. In the back of your mind, there is always something more that you can be doing.

But a lot of deskless jobs really have these healthy boundaries. You made your widgets that day and there are no more widgets you can go make at home. You are done, and you really get to unplug. And there’s a lot of actually positive mental health outcomes for folks that leave desk work and go into deskless work because of that healthy boundary that we have.

So I think a big misconception is that these are always bad jobs. Now, they are bad, though, if they’re not compensated well, if you’re not treated with respect, you’re not given the support to succeed. And too often, companies are undervaluing these jobs, so all three of those factors are there, and they are bad jobs. But I think that that will continue to change.

Jill Finlayson: Yeah, it’s interesting because you point out that they do have the boundaries. In some cases, they can be outdoors, which has also been shown to be very good for your health. Maybe they’re more physically active because we talk about sitting is the new smoking. It’s not good for you to sit all day, and desk workers are certainly doing that. But there are a lot of challenges that are unique to the deskless workers.

Muriel Clauson: Absolutely. Getting back to that point of support on the job, because I think compensation, that one’s pretty straightforward. You have to compensate people at a living wage and/or above it for people to do well in a job. So there’s clear trends that need to change there. I think the population shortage and then the demand for this labor will help to correct some of that, that there’s more pressure on employers to compensate well.

So let’s park compensation for a second. Outside of that, what can we do to support this workforce? Well, what are the supports that help someone be successful on the job? So we’ve been doing research with this population for many years, all the way back to when I was a grad student in my PhD program doing research.

And the most common thing that just blew my mind is how we just make it unnecessarily hard for this population. So I’ll give you an example. We have a company that we work with that had really, really high turnover rates for their first shift at a fulfillment center. And most of the folks at this fulfillment center were Spanish speakers. The people in leadership were not Spanish speakers, so there was really no communication. They didn’t invest in any kind of communication technology to support that. So there was really just announcements given with some basic translation. No two-way conversation ever happening.

Well, there was really high turnover in this first shift. Finally, they figured out a way, which is not too hard with the technology and tools available today, to ask these workers, hey, why is everybody quitting this first shift? And they found out that there literally is just a public transportation schedule that they could have looked into where, for that first shift, workers either have to be over an hour early to that shift or 10 minutes late.

And so for a lot of workers, that extra hour of time they had to invest was just too much, especially for a population that often was working two jobs, had children, other things that they were accountable for. And so that organization just asking the question, finding out they were able to shift the start time for that shift back very little and fix their retention issue.

And I give that example because I think it’s such a good one of the support that this workforce is often asking for is not a drain on the company. It’s actually something that is helpful to the company to the workers that we’ve worked with over many years and thousands of them. They’re not asking for luxuries. They’re not asking for being coddled or anything over the top. They’re asking for just very basic support that helps them be successful at your company.

And often, the biggest gap that we find is communication. And we can get into why communication — I think there’s a bigger conversation there, but communication is the biggest difference between the desk work experience and the deskless work experience, and it has a lot of impact on the support you can receive in your job.

Jill Finlayson: How does a deskless worker communicate with their boss. How did they communicate with the company if their schedule has to change? How does that work today and how should it be working?

Muriel Clauson: Yeah, so they don’t. And it’s not because they don’t want to. Think of your work experience if you’re a desk worker, and think of all the technology that exists to help you get questions answered, access information, understand how to submit requests, share feedback, collaborate, share ideas. You have a suite of software that probably gives your IT team a headache. There is so much available to you and there’s so much enterprise knowledge available to you.

So something as simple as, I want to enroll in the Health insurance that my company offers, you have access to all of the information you need, and it’s still challenging, you still probably sometimes have questions.

Well, what’s the deskless worker experience equivalent in most companies? Well typically, if I’m a deskless worker, I don’t have a computer. I’m not given a company-issued computer. I probably wasn’t given a company-issued email address in over 95% of cases. I might not even have personal email or a personal device like that. And so your whole ecosystem as an organization has been set up in a way that I can’t really access that information.

So what do I have access to? Well, I receive announcements. So I’m given my safety briefing when I come in. There’s flyers, maybe there’s bulletin boards. I’m given announcements of information, but I don’t have a way to ask questions. I don’t have a way to converse. And if I do, it’s maybe an HR team of one on site that’s going to have a line out their office on the lunch break, and I can’t realistically get questions answered.

Not to mention, maybe it’s snowing. I’m wondering if the factory is closed or not before I drop my child off for expensive childcare for the day and take that hit. I don’t actually know the answer to that question until I show up to the factory and see that it’s closed. There are just unnecessary burdens when we don’t think to include this population in communication. And for many companies, we just haven’t thought about how to reach this population.

The amount of turnover that we found in our research that was purely just a worker unable to tell their employer that they were out sick or had to miss a shift is unreal. And that is a cost that companies are constantly trying to figure out how to reduce. So there’s a real gap here that I think is just easily solved if we actually factor in what’s unique to this population.

One of the things we found in research and in the work we’ve done since is for a lot of organizations, if you want to just think about including this population right away, is look for a way to include a text messaging channel. So that’s an inclusive technology. A lot of folks have devices that would work for even in rural areas, which it’s very tough, especially in rural areas for deskless workers where they could, even without a data plan, have access to information at your company. That’s a technology you can absolutely stand up.

And also language is a big one. The deskless workforce is overrepresented by speakers of other languages, meaning they don’t speak the primary language of the country they’re working in. There’s so many technology tools to translate these days and we can solve that gap. But that’s the first biggest one I would focus on, is communication, and we have to get out of our reality of like, we communicate all day. We live in the software that helps us do that all day. What is the experience like for someone that doesn’t have that, and how do we reach those people?

There’s a supply chain Center in Utah we work with in particular that stands out to me, that they have incredibly low turnover. And it’s because they constantly talk about how critical their frontline and deskless workers are to the business. Like they’re constantly saying, like, you’re the closest to our product. Tell us any time that you see anything, we want to learn from you.

The leadership team is constantly saying, hey, we want to learn from you. We want to understand what we can do better as an organization. They include them in planning meetings. They include these workers in development programs. They treat them like they treat their desked workers.

And I think a lot of times for companies, we invest so much in our desk workers — we do our goat yoga retreats with volcanic mud masks. Like we do all these investments, but we haven’t even just consider just opening the door to our deskless workers to be a part of the barbecue.

I talked to so many leaders. I’m like, first tip, include your deskless workers and your all-hands meeting. You don’t have all of your hands in your all-hands meeting. Bring them in there. The more that you’re treating folks with respect, that goes a really, really long way. And then the last pillar is just supporting the workforce to be successful.

If there’s one thing I’ve learned about deskless workers, is they are hardworking people that want to know that they did a good job and that they are going to be able to build a career at your company. That’s a very strong desire. A lot of times deskless workers are multi-generational in their roles.

So my grandfather was a welder. My dad is a welder. I’m a welder. And I’m supporting, as a single mom, my three kids as a welder, and I want to be successful as a welder. These are identity-driven stories they have. They want to be successful, so giving the support to do that.

And one of the ways I see that we can give the support to do that is in our technology initiatives, especially right now, stepping into the AI era, are we thinking about how we include our deskless workers when we think about augmenting roles with technology and supporting the workforce with technology?

Jill Finlayson: I want to break that down, but before we jump to AI, I want to talk about something you just said, which is they want to be successful in their job. And I saw a stat that said 4% make it into a corporate role where 70% want to get into a corporate role. What are we doing to help our workers see the career path, understand the goals or the steps that they have to take to advance in their career? Because that obviously will also help with the engagement and reducing turnover.

Muriel Clauson: Absolutely. And this is an interesting one because — let’s take the US, for example. Like pretty much since the 1930s, we’ve told people that if you want a good life, go to college and be a desk worker, essentially. That’s the good life. And so I think that that message has been felt by this workforce, and there is that desire to break into that world for sure. And I think that that is still the best path to living wages, a good life for this workforce.

Now one interesting thing that we’ve seen is a lot of the workers that, if you break down why they’re interested in moving into desk roles, it’s because they see that as the path to that. They want to have a good wage. That is the way to get there.

I think there’s two solutions here. One is helping to create pathways for folks to move into desk roles if that’s what they want. The other is to appreciate deskless work and give pathways for folks to be successful and to meet their financial goals and their career goals within a deskless job as well. I don’t think the solution is that we help 100% of our deskless workers become desk workers. I think the solution is that we value and create career paths on both of those trajectories.

So let’s take desk work for a moment because there’s been a lot of really great headway that’s been made across the US and some other countries to remove some college degree requirements to move into desk jobs. I think that is an incredible initiative. I think every organization should ask themselves, is it really necessary to have a college degree if you’ve demonstrated competencies for this role working at our company in another capacity? So that, I think, is just table stakes to do that.

On top of that, one of the best ways you can actually begin the upskilling process is just exposure for the workforce. So for a lot of deskless workers, they are related to deskless workers, their friends are deskless workers. They actually might not even understand what some of those desk pathways are.

And so giving the ability to have a glimpse into what those jobs are, getting to know people in those jobs in the organization, understanding what they do, especially if those are role models that they kind see themselves in, you can start to get folks to have even that vision for what is possible.

And then additionally, there’s incredible programs that have focused on how do we actually upskill a deskless workers so they’re getting the skills they need today, but also moving into those pathways?

Jill Finlayson: Those are important ways for people to grow in their career, and I like the fact that you differentiated for those that want to go into desk jobs, here’s one pathway, but for those who just want to advance in their technical role and have advancement opportunities, there’s another pathway. I think this relates back to your point about communication devices.

So if we want people to be upskilled, we need to be able to deliver that upskilling. And the other thing I wanted to throw out there and then just get your reflections is, can they get upskilled on the job while they’re being paid or do they have to do it on their own time? Because that’s another barrier to participation.

Muriel Clauson: There’s interesting examples of different companies that have tried different things. We work with a very old manufacturing company. The Chief People Officer is such a great example of thinking about really meeting people where they are and has done a great job with their deskless workforce.

What she decided to do is she heard that some of the deskless workers were saying, hey, all these workers get to work from home. We’ve never had that opportunity. And she decided to actually get laptops for their deskless workforce and compensate them for some work-from-home time where they actually were taking the trainings they would have been taking in a room on-site on laptops at home. And that went a long way for retention in that organization and really helped.

We do see a lot of success, though, with on-the-job training. And it also does actually raise an individual’s buy-in even for the work that they’re doing when there is a learning component to what they do. That’s like a very — post-onboarding any kind of learning initiative with your job., it raises your esteem for that role in general and across deskless workers in particular.

And so helping folks learn on the job, we’ve seen really cool examples of workers even designing their own approaches to training with generative AI solutions. Like, that is important. I think the biggest barrier to it is that we don’t have technology to deliver training, learning opportunities to this workforce in general.

Most organizations we’ve seen, they still have the room with the projector, and they are just getting the most boring presentation. I think thinking about, OK, how are we investing in learning and development across our desk workers? How do we actually take a similar approach across our deskless workforce is really powerful.

One other piece to this, though, that I think is really important is if we talk about just advancing in a deskless pathway, that, honestly, for a lot of companies we talked to, has to be the goal because they have a real shortage across this population.

So part of that is thinking about different compensation tiers, but another part of this is actually thinking, well, how are we investing in technology that’s going to augment certain parts of this work, and then how are we helping workers adapt to that and grow alongside the technology so that they’re able to use it?

And so a lot of the companies we work with, when they’re thinking about things like generative AI, like, this is actually the population that we say you have to look at first. It’s not the population to look at last because this is the population that it has the biggest shortage that you literally will not be able to staff the operation fully, so you need augmenting technology to support this workforce.

Jill Finlayson: So tell me, then, what is the use case for using AI, generative AI, for the dustless worker?

Muriel Clauson: So there’s so many, and I’ll take a step back for a second because if you remember in 2013, there was a working paper that came out, Frey and Osborne out of Oxford where they were miscited by pretty much every newspaper in the world to say the robots are going to take our jobs.

They did some work where they did a Bayesian analysis where they basically looked at jobs that exist today, capabilities of AI, and then which jobs have the potential to be automated. And their work was totally miscited and turned into this big conversation around, oh my goodness, the robots are going to take our jobs.

And that work especially made people think, well, the robots are going to take deskless jobs. Like, these are the jobs that are going away first. And you still see the ripple effects of young people opting into things like manufacturing based on our misinterpretation of what we talked about in 2013. It’s really interesting.

But what we learned following on from that research — and there’s been several iterations — I wrote the textbook through Cambridge where academics write reviews of a lot of research, and basically when we were doing that work preparing this and looking at all the studies around job automation, we found that what has been really the most accurate view of how technology changes work, particularly deskless work, is this idea of augmentation.

So historically if you look at automation, replacing a job, automation typically actually leads to total worker growth. So we’ll see what’s called the capitalization effect. The more you invest in replacement technology, the more net growth you have across your workforce. It’s an interesting concept. You can look into it. It’s called the capitalization effect if you’re interested.

So that replacement theory didn’t hold up at all, and particularly not with the deskless workforce. What did was this idea of augmentation. It’s the opposite philosophy. Instead of saying, OK, well, what’s technology good at and how do we use that to replace a worker, we instead say, what is this worker doing that is hard or repetitive or dangerous or just not as effective as it could be? And is there technology that can augment and help this worker? So it’s really just flipping it upside-down from how we thought about it before.

We have seen augmentation have a profound impact on especially deskless workers on many kinds of fronts. So safety is a huge one. There are just safety risks that exist across deskless jobs that don’t exist for a lot of desk workers. That is an incredible place to invest to make these jobs better, to make sure you can actually continue to fill these positions.

We’ve seen incredible examples of learning being accelerated. So we have a manufacturing company we work with trying to do a transition from basic manufacturing machines on the floor to advanced manufacturing. And one of the workers said that he played with ChatGPT.

He’s like, I think that what’s possible is actually if we took this whole training content that you developed, we took the manual for the machines, we took the OSHA safety training requirements, and we ingested them into our company’s version of ChatGPT, each of us workers could actually ask it questions and learn as we go on the machines. And this idea was so effective that the training that was supposed to take 18 months for this transition ended up only taking a few months. And so it accelerated the learning of this workforce.

There’s also just efficiency gains that can come for workers, which workers want. There’s not this idea of, oh, if I’m more productive, I get replaced. They’re seeing the shortage on site and they’re seeing what is possible. So there’s a lot of great ideas.

One thing I’m really bullish on, though, is that the executives at the company should not have the ideas, the workers should. So workers often have the best ideas about how technology can support their jobs. We just don’t ask them, we don’t talk to them. We don’t give them that opportunity, but workers know what’s hard about a job. They know best how to augment a job, and augmentation works better than automation any day.

Jill Finlayson: That’s an important thing, that the frontline staff are engaged with the problems. They understand the issues, they understand where their biggest pain point is. And so if you can give them access to the tools, they’ll figure out how to solve it for themselves. So what do we have to do in order to give them to access to these tools? Because as you point out, they don’t even have basic communication, how are we giving them access to AI?

Muriel Clauson: This is like where I’ve been investing most of my efforts over especially the last couple of years because I think that this is the piece we have to get right to actually include everyone in the AI era.

You found the thing that I get really passionate about, so here we go. The leader’s role in an organization is to create a climate of AI readiness, not to come up with use cases. I don’t think leaders and companies should be coming up with use cases. So what does AI readiness mean?

Well, AI is only as impactful as basically the data that you use within it. And for companies, especially for trying to augment work, you have really great historical, structured and unstructured data that can help in learning initiatives, safety initiatives, productivity initiatives, all of these different things this technology can help with.

And we’re talking right now specifically about generative AI. A lot of us have tried things like ChatGPT. Basically what companies have the ability to do today is use something called Retrieval-Augmented Generation. So that is basically a RAG for short.

And what RAG does is it basically takes the power of a large language model to generate text to sound human, all of those good things, but then layers in a constraint around, well, what is true in the model? What is the data that we’re going to rely upon? So instead of relying on the internet, you rely on the enterprise data, the enterprise knowledge that you’ve put into the model. And so if you do that, you’re actually able to give every worker access to all of the data that you put in there to support them in their job.

Now, why is that different than things that we’ve done before? Well, the main difference is that how do you actually access that and make use of it? Well, you don’t need special software. You don’t need special coding skills. You don’t need to design all of these point solutions. What you need is just a conversational interface.

So if you open up to your worker — let’s take the manufacturing company learning example. If you give them access to Slack, Teams, a widget, maybe a text message where they can just write a sentence, that worker can design their own AI tool as long as your enterprise knowledge is inside basically your RAG model.

And so because it’s that simple to work on those use cases, I think the job of a leader, the job of the IT team, the technical team, the executive team, is to say, OK, let’s get our security in place. Let’s make sure we have our data in here and we have the right data quality.

Let’s make sure that we have observability of how this is used. And then let’s have the right adoption channels. Let’s make sure every worker in our organization has a conversational interface where they can essentially query this data. And there’s ways to control permissions and make sure data is in going places it shouldn’t.

But if you do those four parts, you’ve set up a culture where you don’t have to decide as executives what use cases you’ll have around AI to support workers, every worker can actually design their own augmentation tool.

If you’re interested in learning more about those four parts, we did write a whitepaper about this. You can go really deep on it, and I can give information in the show notes.

Jill Finlayson: And not to be too literal, but what do they access this on? What has to happen so they have that conversational interface?

Muriel Clauson: That’s one of the big investments you need to make of the four. Everyone needs an adoption interface. So some place they can type a sentence. In manufacturing plants, we’ve seen people add widgets to machines. So web widgets to machines.

We’ve seen companies expand their Teams or Slack licenses to include their deskless workforce. We’ve seen some companies that, yes, choose text message, and yes, you can do this via text message in a more limited way. We’ve seen installations of things like iPads on the shop floor and all of that.

If you want to see your investment pay off tenfold and more because of what is possible, make sure every worker in your company has a conversational interface where they can type a sentence if you want to include everyone in the AI era in your company. So practically yes, that is like a big gap that you have to fill, I believe, as an employer.

Jill Finlayson: So let’s take it up one level. What should governments be doing?

Muriel Clauson: You know, my background was working with governments on strategies. Early on — and we were having conversations around this, I remember in 2015 where we were talking about generative AI in particular, AI is a huge world. Generative AI is one tiny corner of that world. And if you’ve played with something like ChatGPT, you’ve tried one tiny corner of the gen AI world. It’s a big realm.

But generative AI is unique in the sense that because it’s so accessible from a language perspective, you can just write a sentence and mold the tool to do what you want it to do. This is really an opportunity, I think, to make technology more inclusive than we ever have.

So you don’t necessarily need a ton of special skills to access it. You don’t need to know how to use a mouse, click around different icons on a screen. You don’t need to have skills even on how to use the internet or anything like that. You really can use it if you know how to type a sentence — and even type a sentence with poor grammar or maybe in a language that wasn’t the preferred language of your organization. This technology really helps to detect what you intend from a sentence.

So in the context of gen AI, the most important thing for policymakers to think about, I believe, is human in the loop. Because what we already talked about is, OK, we have these large language models that basically are really good at replicating human dialogue. They’re good at creating content, sifting through how we write things to help us retrieve information.

We then use RAG on top of it, Retrieval-Augmented Generation, to help us retrieve the right information that constrains what is true. So in organizations, that can be data that exists in the company. And then we have our semantic layer on top of how do we govern that data, which, if you want to read more about those four parts, you can read the whitepaper.

So from a policy perspective, where do we really, really have to get it right? Where we have to get it right is human-in-the-loop approval of basically how that data is ingested and used. So that really critical piece is data quality. When we think about enterprise knowledge, what’s true in a model, I think one of the most important things is that you have a person that is verifying that the knowledge that’s in there, that the truth that’s in there is correct, that it’s in the right form, that it will be used for the right things, and that it will go to the right people for those use cases.

And that is something that I believe, from a policy perspective — and the EU AI act did a great job of addressing this, we have to require of our tools. I think if we give up that piece, that is where we’re going to see a lot of problems.

We have to know that when we’re giving access to data, that we’re giving people this support, that that’s information we can trust, and I think that that is the big piece that we have to get right from a policy perspective. So the human-in-the-loop pieces.

I do believe from a policy perspective, the EU AI Act does a really great job of setting the standard. I encourage every company we work with in whatever country to abide by it. We require it on our platform, that you abide by the EU AI Act because I’m a big supporter of it. If I had to pick just one piece, human-in-the-loop data quality approval has to be there.

Jill Finlayson: And something we’ve seen on the desked worker side is they’re starting to use ChatGPT whether the company has given them permission or not, sort of shadow IT. And so are you seeing this with the deskless workers? Are they jumping into this on their own time already?

Muriel Clauson: Yes. So there has been a pull for this technology like we’ve never seen before. So usually, I mean, think of SaaS, Software as a Service, solutions. Somebody in leadership decides which SaaS platform to buy, and then you disseminate it down to the workforce and the workforce uses it. That has flipped on its head.

The biggest adopter body within companies of generative AI solutions has just been individuals throughout the company. If you look at most of the adoption reports — MIT did a great one with Databricks, you basically see that there is so much more pull than push of this technology.

So what have companies done well? Most we talked to, they’re like, we are just putting up firewalls, basically, to protect against any of these tools being used by our workforce, and there’s this just pause being thrown down. And you’ll understand. I mean, companies can have their data ingested by external tools that they don’t know what they are, and all of that makes sense. It’s really valuable that you maintain that data.

But I think that that’s a really good sign for organizations that you have to be like, great, our workers have ideas of how to use this. They want to use this. Let’s set up the AI readiness that we need so that they can do this in a way that we can approve. And so I love talking to leaders that are at that firewall juncture and ready to move to the next step and empower the workers. But how cool that the workers have the ideas.

As we talked about even back in 2015, I think this is the most inclusive technology that we’ve seen in a really long time, and I see that’s playing out with the workforce and all workers.

Jill Finlayson: So let’s talk about the worker and what they can do because if their supervisors or their company are not creating this enabling environment, what are the options for them and how should they be interacting with generative AI?

Muriel Clauson: So one of the things we tell individual workers is like, if you have a really good idea on how to use generative AI to help augment your work — so to help you do more, better, faster, more effective, learn, tell your organization because most executives that we talk to do have a bit of a crisis of imagination.

Most companies we go into, we meet with their executive team, they’re like, we have a really good idea of how to use generative AI — customer service. And we’re like, OK, cool. What else do you have? What else do you have? Like, that is really where a lot of companies are at. Like, they really have just scratched the surface of what’s possible.

So if they have workers that have ideas that are excited to use it, share those ideas because most executive teams we work with are trying to find the opportunities of how to augment work and not coming up with, I think, as rich of ideas as we hear that the workforce is.

Jill Finlayson: Given the power dynamics, and in some cases, as you were saying, the language dynamics, how can they start to raise these issues if they don’t feel empowered to do so?

Muriel Clauson: I put a lot of onus on the leaders to do that. If you do have a power dynamic where people can’t share and communication and language barriers where people can’t share, you don’t get to hear these ideas. So that’s actually one of the first things that we say before we’re even willing to work with a company, is we’re like, we’re not going to help you get ready if there’s not a way for your workers to drive use cases, let’s set that up first.

That’s often, I would say, step one in the process, is how do we learn from our workers on the use case front? Because you’re right, if someone’s not empowered, you’re never going to hear. And what a bummer for all these companies. They’ve got all these brilliant people across their deskless workforce that they are never going to hear any of their ideas because they speak a different language or don’t have communication software. What a bummer.

Jill Finlayson: And what a missed opportunity to be able to innovate and be competitive and retain the workforce that they have.

Muriel Clauson: And when there’s one deskless worker for every eight jobs, they’re just not going to work at your company if you’re not a company that is empowering them with these tools. They’re just not.

Jill Finlayson: So what is the future of work look like for the deskless worker?

Muriel Clauson: I think really bright. There’s these interesting flex points throughout history of work that I think we’re in right now. I think that there is a supply-and-demand shift that is happening around deskless work where these are going to actually be some very celebrated roles. These are going to be well-compensated roles.

There’s going to be roles where we treat people well. I think we’re in that messy transition point where companies are like, what’s going on? And workers still don’t know how much they can ask for. But I really see that we’re coming into an era where we celebrate deskless work much more than we do today and compensate well for it.

I’m very optimistic about these types of jobs. I think if you have a child that is interested in going into this type of work, I would encourage that. I would support that. I think there’s a really good era for this worker coming.

Jill Finlayson: So what are your final words of advice for our listeners about how to advance equity and inclusion for deskless workers? It’s 80% of the workers. It’s a huge percentage. How can we enable them to be more effective in championing their needs?

Muriel Clauson: So first and foremost, acknowledge that they exist. It is so funny how many times we talk to leaders that are like, oh, yeah, we haven’t included these people in anything. So just having that realization, that moment of humility of like, wow, we’ve really dropped the ball here I think is step one.

The second one, and we’ve talked about it a lot, is you have to open up a communication channel for reasons we talked about around just basic retention, basic respect, basic relaying of information for your operations to run smoothly. But now, stepping into the AI era where we have generative AI is only available to people that have a conversational interface that they can write a sentence in. So communication channels have never been more important. We have to include all workers in this.

And then third, and I think this is where it starts getting fun, is actually listen to these workers. If it’s a service-type role, they’re the closest to your customer of any worker in your organization. If it’s a goods production or movement-type role, they’re the closest to your product of any person in your organization. These are incredibly important workers, they have great ideas. Just wait if you open up the channels to actually learn from them.

Jill Finlayson: I think even Elon Musk said humans are underrated. They can really problem-solve, they can really get out there, and be the engine to drive your business into the future and drive their own career advancement into the future.

Muriel Clauson: Absolutely. And that was a manufacturing context, which is often, I think, the first place we look to replace workers. And it just hasn’t borne out. Augmenting workers is the way to go. Supporting workers to do more is the way to go, and absolutely, humans are underrated.

Jill Finlayson: Amazing. Thank you so very much for all of your insights and really appreciate you opening up this topic and creating the dialogue that’s needed to be able to move the dial forward in inclusion.

Muriel Clauson: Thank you so much for having me. I really appreciate it, Jill.

Jill Finlayson: And with that, I hope you’ve enjoyed this latest in a long series of podcasts that we’ll be sending your way every month. Please share with friends and colleagues who may be interested in taking this future of work journey with us. And make sure to check out extension.berkeley.edu to find a variety of courses to help you thrive in this new working landscape. And to see what’s coming up at EDGE in Tech, go ahead and visit edge.berkeley.edu.

Thanks so much for listening. And we’ll be back next month with another look at the future of work. The Future of Work Podcast is hosted by Jill Finlayson produced by Sarah Benzuly, and edited by Matt DiPietro and Natalie Newman.

--

--

UC Berkeley Extension
UC Berkeley Extension

Written by UC Berkeley Extension

UC Berkeley Extension is the continuing education branch of the University of California, Berkeley. We empower learners to meet educational and career goals.

No responses yet