JAIME:
As we’re thinking about the future of work, even in the context of AI, you know, we talk a lot about how we can start automating the drudgery or, sort of, the repetitive parts of work. Sometimes those pieces of work are actually important for human attention. It’s not just about like, okay, raw, what do the large language models do well, how do we bring them together to make it better? But it’s like, how do we set people up to contribute the best, to think well, to see things in new ways?
REID:
Hi, I’m Reid Hoffman.
ARIA:
And I’m Aria Finger.
REID:
We want to know what happens if, in the future, everything breaks humanity’s way.
ARIA:
We’re speaking with visionaries in every field, from climate science to criminal justice, and from entertainment to education.
REID:
These conversations also feature another kind of guest, GPT-4, OpenAI’s latest and most powerful language model to date. Each episode will have a companion story, which we’ve generated with GPT-4 to spark discussion. You can find these stories down in the show notes.
ARIA:
In each episode, we seek out the brightest version of the future and learn what it’ll take to get there.
REID:
This is Possible.
Obviously, you know, 2023 being the year of AI, one of the primary things that everyone’s talking about is: what is the future work? That’s why we’re so excited to be talking to Jaime Teevan. Jaime is Chief Scientist and Technical Fellow at Microsoft, where she was responsible for driving research-backed innovation in the company’s core products. Jaime is an advocate for finding smarter ways for people to make the most of their time. She leads Microsoft’s Future of Work initiative, which explores how everything from AI to hybrid work changes the way people get things done. And part of the reason, of course, you know, doing the book Impromptu was to say, “actually, in fact, we’re going to have these ‘aha!’ moments,” e.g., the amplification of human ability where AI is amplification intelligence versus, you know, kind of, artificial intelligence. Because people are always talking about replacement versus amplification and augmentation.
And I think Jaime is going to be great to talk to about this because, you know, part of her research for years, the things she’s been doing at Microsoft, I’ve talked to her a number of times where she’s bringing a kind of a clear, kind of a scientist, and a, “let’s look at the data, let’s study this,” about, like, “what are the things we could do to help people work? What are the things that make them more productive, happier, you know, more connected, more creative in the things that they’re doing?” And so, this discussion about the future of work will be kind of grounded in real lenses of the future versus, you know, what you normally get, which is just people’s fears or uncertainties as they approach this topic.
ARIA:
And I think everyone is, of course, so interested in this topic because it’s so personal. Where you work and what you do. Everyone says, you know, you spend 40, 50, 60 hours a week doing this work. And so, what is it going to look like in 10 years? What is it going to look like in 20 years? Like, that’s what we all want to know. And I think also personal productivity, it’s like, how can we all be better at our jobs?
I’m also so excited about this episode because the future of work affects everyone. So, Reid and I reached out to a few key people in our network to hear their hottest takes about what’s to come in the world of work. You’re going to hear voicemails from these special guest stars throughout the episode.
Here is our conversation with Jaime Teevan.
REID:
Jaime, it’s a great pleasure to be doing this. One of the things that I have learned from our years of working together in the Microsoft thing is how thoughtful and, kind of, data- and truth-oriented you are in these things. It isn’t just kind of like an evangelism, it’s a, “no, no, we’re studying how to make this stuff better because it’s not just a question of opinion – is it going to be like good, bad, ugly, you know, wonderful, you know, da da da…” But actually, “no, no, here’s what we’re learning and here’s how we’re proceeding very intelligently.” So, welcome to the Possible podcast.
JAIME:
Thank you, it’s my pleasure.
ARIA:
Jaime, it’s so lovely to meet you. I was saying earlier that I think of myself as sort of like a superhero because I’m the mom of 3 boys, but then I read that you are the mom of 4 boys. [laugh] And I was like, “oh my God, I need to learn everything from this woman.” And then actually that, that moves into my first question. It’s like, everyone has to, but especially moms, dads working parents, it’s like we’re on this constant quest for productivity. Like, how can we be more productive? How can we leave more time for the things we want to do? And you’ve done so much with micro-productivity and microtasks. Do you think microtasks are the way of the future? I would love to hear your thoughts on them.
JAIME:
Oh, that’s such a good question. And actually, my kids, I strongly believe, have made me more productive. Like, there’s this forcing function that moms have, like in parents, just you have to use your time and it forces you to use your time thoughtfully. Like when I’m working, I’m focused on work and I’m not going to screw around at work because I know I have other things that I could be doing if I’m not. And I would say, you know, we’re talking a lot about sort of how work is changing and in particular how technology is going to change work. Yeah, it was interesting for me even just to reflect on how much has changed since my kids were born. So, Griffin was born in 2004 and when he was born, Facebook didn’t exist, Twitter didn’t exist, the iPhone didn’t exist. Some important things that really have opened up a lot of the AI revolution too, like ImageNet, didn’t exist, crowdsourcing wasn’t a thing. And so much has changed.
And I would say the research that I did related to micro-productivity was really inspired by those technological transitions in addition to, sort of, my family and personal transition. Here I am, I’ve got 4 babies and it’s overwhelming and it’s so much work, because I never know when somebody’s going to wake up or need something from me. And so I got really interested in how we could use the phone, use these small little bits of work that we, I mean, in some ways, what you do on Facebook or Twitter or even in crowdsourcing, are little bits of work. And we were getting, you know, as a scientific community, we were getting very smart about how to take these little bits of work and stitch them together into something bigger.
And I got interested in how I could do that for myself so that while they were napping or doing something else, I could be doing something that was productive and valuable to me.
ARIA:
And I feel like so often we think about that as interrupting our work. It’s like, “oh, you can only do work if you need 3 hours of time to, like, sit at a keyboard and do 3 hours in a row.” But you found that we never have that. Can you be productive in these short bursts? And what did you find about that?
JAIME:
There’s value to focused work and taking a lot of time, so I’m not trying to discount the value of that work, but there’s also value to these small little bits and it can even be intentional. So we’ve done some research, for example, that shows coming at the same problem from different perspectives, or at different times, or different locations actually inspires creativity. So you may have an email that you want to respond to, and it’s hard, and eventually – you know, like, this is a comment I get emails and I’m like, “oh my gosh, what a jerk, why are they sending me this annoying mail?” And we all know that you shouldn’t reply to that mail immediately. [laugh] And that ability to sort of look at the mail, internalize the mail, think about what matters, think about the different kinds of points you want, write pieces of it, and come back to it actually allows you to internalize the content that you receive, see what’s valuable, you know, to get over, sort of, your initial rage reaction or response and and see what’s valuable there and then respond to it in a meaningful way. So there’s all sorts of different things that you can do with these little bits of time.
REID:
Say a little bit about how these, like, microtasks, and the work is evolving, given this M365 Copilot, and what the work that you’ve done around, like, how it gets crafted to really be helpful and really on target for making people more creative and productive.
NEWS REEL:
Alright, so Microsoft announced a new feature called Copilot that integrates AI technology into its Microsoft office software.
NEWS REEL:
Copilot combines the power of large language models with your data in the Microsoft graph and the Microsoft 365 apps to turn your words into the most powerful productivity tool on the planet.
NEWS REEL:
Microsoft’s promise is that this will effectively allow you to start interacting, via natural language, with your computer.
JAIME:
A lot of the early micro-productivity work was sort of looking forward to this moment where AI was able to be a real participant in what you’re doing. And thinking about that, one of the things that’s happening with things like the Microsoft 365 Copilot is that we’re now able to use language to engage with the system, but more than just using language, we’re able to use conversations. So, it’s like that back and forth and that ability to iterate. And so, when we were just talking about what it means to look at something from a new perspective, it becomes really useful if the system can actually be helping in coming up with different ideas and different perspectives, and then you have the space to respond to that.
ARIA:
It’s so funny, just today I just got an email from a friend who – she has a one-woman social media agency, and she was saying that she’s been using ChatGPT to be her partner. She’s like, “I’m only one woman, and so ChatGPT gives me feedback and gives me ideas and I bounce ideas off of it.” And it was so interesting that, you know, she was saying, “this is my coworker for the time being because I’m starting out, it’s a startup, and it’s just me.” And to your point, like that’s how she’s using it as her copilot.
JAIME:
When you think about language generation, one of the obvious things is content creation. Of course you can say, “oh I have some ideas, make it into a lot of content.” Another place where it’s sort of obvious is summarization. How do you take a piece of content and then summarize it? But exactly as you’re saying, Aria, it can do so much more than that, and it becomes really interesting when it can give you feedback.
So I’ve actually found that when I’ve had, for example, hard emails to write or something that I’m not exactly sure how it’s going to land – and I would guess that this is great for somebody who’s a social media person as well – you want to see the different perspectives people might have. Like, what are the things that might raise red flags for others in this so that I can respond to that first?
JESSI:
Hey Aria and Reid, this is Jessi Hempel. I’m host of LinkedIn’s flagship podcast, which is called Hello Monday. And at Hello Monday, we spend our time thinking and talking about the future of work specifically, and how that work is changing us. And so, you asked, “what is your hottest take about how we will work in the future?” And I have one very clear thought about that. The work of the future, for us, is relationships. This becomes so much more clear as AI comes on the scene and takes over so many of the aspects of what I had always thought of as work, work that we needed specific education to prepare for. Now AI can do it in minutes, right? But the thing that uniquely is ours, and the thing that we’ll need to lean into as all of the institutions around us get reinvented, as all of the norms we have come to depend on are reexamined, is the ways in which we can be present with each other, in which we can support each other, in which we can know each other. Our relationships.
REID:
You know, some of your work a couple years ago was also on what was going on with hybrid, and COVID, and the pandemic. Microsoft released a report on how remote work affects collaboration. What do you think were the key learnings, going through the pandemic, relative to hybrid teams, hybrid work, remote? You know, what should people have learned from that, going forward?
JAIME:
Yeah, and it’s fascinating how intertwined – even if we maybe think of these two major disruptions to work that we’re experiencing right now: one, the rapid shift to remote and then hybrid work, and the other, AI – how much they’re actually intertwined and related. You know, in many ways, I think of the shift to remote work as breaking down spatial boundaries. So, all of a sudden, space was something that we stopped thinking about in the same way. I can work remotely, I can collaborate across different time zones, across different locations. But in the process, that broke down a bunch of other boundaries because space actually affords temporal boundaries. And when you go into the office, that’s when work starts. And when you go home, that’s when it ends. It broke down a lot of our boundaries between work and life. It forced us to rethink the technology of space in a fundamentally different way.
It’s basically a technology we’ve been using for millennia to get things done. You know, if you want to transact, you go in person. If you want to bump into people, or have a brainstorm, or have new conversations, you go and do that with them, physically, co-present. And now we’re thinking about how you can do that in a technologically mediated way. The fact that so many conversations are now technologically intermediated, AI now has this real opportunity to help us figure out, like, we sort of threw up in the air: what does time mean, what does space mean? And we need to make sense of that and we have all this new data and all these new surfaces on which people are interacting, and we have this opportunity then to bring AI to help make sense of that and figure out where it’s going to go.
REID:
Well I’m looking forward to when we’re doing, you know, kind of remote podcasts like this one, because one of the things it does enable us to do is relatively easily assemble, with Aria in New York, you and I in Washington, you know, but in different locations to assemble us.
But to have the AI copilot going, “oh, you just said this, ask her about this thing!” [laugh] Right? I’m looking forward to that.
JAIME:
You don’t have long to look forward, because we actually, I mean, we now, in [Microsoft] Teams, can actually help you real-time in a conversation and especially be like, “oh wait, Reid, you meant to talk about this!” Or, “you seem like you maybe disagreed about this topic!” And we can start tackling that as well.
ARIA:
Talking about, like, what the difference in work is going to be in 5, 10, 20 years, I remember when I joined the workforce force, my boss had a big paper calendar on her desk and the only way you could get time on her calendar was by going into her office and writing, you know, 4:00 PM today, and then we erased it if something changed. So, the idea of paper calendar seems really quaint now. What are the things that we’re doing today that you think, in the future, are going to be quaint? Because we’re just, we’re not going to do them. It’s just going to be a different world of work, if we look down the road.
JAIME:
Certainly the idea of not being able to speak to your computer will be weird. [laugh] I suspect the kind of artifacts that we use for communication are going to change in a fundamental way, as well, with conversations becoming where knowledge is embedded. And so, really our focus is going to be on how do we talk, how do we communicate, less on how do I put this in a document and present it separately.
ARIA:
Let’s turn to, to what GPT-4 posited, that the world of work might look like in the future. So the first AI story, it was about Anna. She was a 35 year old woman in Mexico City and she uploaded all of her proprietary data and she created LLM clones of herself.
AI STORY:
Anna enjoyed her work at her startup, Lingo, but she also had a personal project that she pursued in her spare time. She wanted to create an LLM clone of herself that could act as her alter ego in the virtual world. She saw it as a way of expressing herself, expanding her horizons, and as a potential source of income since she could use it to take on more work opportunities without sacrificing her own time and energy.
She started by collecting and organizing all the data that represented her identity and experience. Her projects, notes, intellectual property, social media posts, emails, chats, photos, videos, audio recordings, biometric data, and more. She stored, encrypted, and backed up her data to ensure that she had full control over it. She then used Lingo’s LLM as a base model and fine tuned it with her own data, using 100 gigabytes of cloud storage space. She also added some features like emotion recognition, voice synthesis, face generation, and personality profiling.
ARIA:
The AI story posited that instead of doing the work, she actually became a manager of 4 or 4 different “selves” of her that were doing the work for her. So, just, I would love your reflection on that story. What did it get right? What did it get totally wrong? Like, where do you see that as part of the future of work?
JAIME:
It was a fun story. Probably the biggest challenge I saw with it was it really personified the AI. It was: how does AI become a replacement for Anna? And I think it’s much more interesting to think about how it can make Anna better or do things differently. So I actually used GPT-4 to ask some questions about-
ARIA:
I love it! [laugh]
JAIME:
-the story. And, you know, the first one was really to help me think through it. Actually, the first thing I did was – and I’ve been doing this for all my documents, you know – you go to all these meetings and you get pre-read documents. My favorite thing to do when I get a pre-read document now is to get it summarized, but not just raw summarized. I like it summarized as a poem. For some reason, it’s a lot more fun for me to read it as a poem and brings a little bit of joy into that interaction. I actually pay attention and I’m like, “oh, that rhyme didn’t work!” So I summarized it as a poem, but the thing I liked best was actually encouraging the model to help me think of ways to think further. So I gave it the prompt. I’ll read the prompt, I said, “as good as LLMs are, I don’t think using them to clone people is the right thing to do, it doesn’t seem like the best way to capture their potential. So, what other things might Anna have done with the technology that are really different and novel ways of imagining it?”
And it had a bunch of interesting suggestions. One that I really liked was thinking about it as a digital mentor/coach to help her grow and think differently. Another one was to think about using the LLM as a digital artist or storyteller. And, like, you know, I think of all the work that we do just to capture our lives and think things through, and how special it would be to have that captured as well.
ARIA:
No, I think about the coach all the time, because I feel like in so many organizations, you know, the CEO, the executive team, of course they have a coach, of course they’re talking to someone outside to make them better. And, you know, we would love to have a human coach for everyone in the organization, but it’s not going to happen. And so, if we can have an AI coach to help everyone at every level, I mean, we’ve just unlocked so much productivity. So that’s awesome, I love that it came up with it.
JAIME:
It’s a little bit of, like, then she creates a manager to manage her clones that she creates, and it’s a little bit turtles all the way down. So, I actually asked about that as well, and it came up with sort of this story of imagining clones of clones, and I thought that was sort of a fun one. It says, “Anna created an LLM clone of herself to work for 4 different employers at the same time. And she was so successful that she decided to create another clone of herself to manage her clones. And then she was so busy that she decided to create another clone of herself to enjoy her life. And she was so happy that then she decided to create another clone to share her happiness. And then she was so cloned that she decided to create another clone to help herself remember who she was.”
REID:
Yes, and now the phrase won’t be “turtles all the way down,” it’ll be, “clones all the way down.”
ARIA:
Clones all the way down. [laugh]
REID:
I think there’s a couple of very good highlights. One is that it’s a copilot. And it doesn’t mean that what it says is necessarily true, it doesn’t mean that what it says is the complete story. It’s a way of, kind of, elevating your game. I also, when I read it, I was like, I don’t think the clones thing is particularly useful. And the real thing is how are we, you know, how is this tool really catalytic? You know, it can help with the blank page problem. I don’t think we’re going to get to – and we’re doing a possible podcast – we’ll say, “well, my clone will talk to your clone!” [laugh] You know, and that’ll be done. But it can help us elevate our game.
PAPIA:
Hi everyone, my name is Papia Debroy, and I’m with the organization Opportunity@Work. To make our communities and citizens competitive and resilient to what are now constant changes in our economy as the future of work is evolving, we have to activate the potential of the entire US workforce. That includes more than 60 million workers who have a bachelor’s degree. It also includes the more than 70 million workers who are skilled through alternative routes, or “STARs.” And they’ve faced incredible misperceptions. The majority of managers in this country believe the majority of the workforce has a bachelor’s degree. This impacts their perceptions of who should get jobs. It also impacts how managers invest in their current workforce.
These are invisible barriers. They come at every turn for workers without bachelor’s degrees. Yet our analysis actually suggests more than 30 million STARs have the skills to make transitions to jobs that pay on average over 50% more than what they currently earn. It’s time we rid ourselves of that invisible barrier, that we tear the paper ceiling and see the world beyond it.
REID:
Before we go to some more of the AI stories, I think one of the things we should address is, obviously, the general dialogue around AI, generally, in the press is a worry about job replacement. What does your research currently lead you to think about how jobs are going to be transformed? What is, like, you know, a lot of people listening to this will have like, “what about my job?” You know, like, I even go to an investor conference and I have investors ask and be like, “well, what is AI going to mean for my job?” [laugh]And you’re like, “well, you’re going to be able to invest in a lot of really cool AI things! That’s one!” [laugh] Right? But anyway, so what’s your current lens on this and how would you talk to the, kind of, the general dialogue on this?
JAIME:
Yeah, no, and it’s terrifying, it’s really scary to have this new powerful AI tool to work with. I get scared, too. Like I have a lot of empathy for that. These large language models are a really, really powerful tool, and we’re going to have to figure out how to use that tool. And it’s going to open up a lot of opportunities. It takes time for our imagination to really get to work and figure that out. You know, anytime you think about a new technology, sort of the most obvious uses of that technology are substitutionary. Like, I even think about something very simple: think of GPS technology and how we initially used GPS technology, like, “oh, it’s great, it’s a map, I don’t need to know where I’m going, I can just follow my phone and it will tell me where I’m where I’m going to go.” And it’s been great that way, and that has changed. I no longer need maps. I actually don’t pay as much attention to where I’m going. People do a little worse with navigating and locations.
But what we, you know, what we didn’t see at the time, what was hard, is all of the new things that it was going to bring. I can now quickly find places to eat near me. I can find, I can know where my husband is on his way home. But then we can start looking at that data and start thinking even more creatively. We can see traffic because we know where people are traveling, we can do better road planning as a result of that. And so, all of these complementary uses of the technology are going to emerge and are starting to emerge, but they’re hard to see. So, I think it makes sense that we find it scary, but there’s a lot of opportunity to do things that are pretty amusing and interesting.
REID:
There’s a relatively smaller number of jobs that’ll just be completely eliminated. Not zero, but completely eliminated or reduced substantially. I think which jobs will be transformed, so, what the work you were doing before, you know, now or before, will be different than the work you will be doing. Like, one of my big hopes is as a lot of the equivalent form entry part of things will now be all much easier, and people can focus on other things than the form entry part of their jobs. Because we almost all have some form of form entry as part of our job, some people more, some people less. But, like, one of the other AI stories we talked about was, you know, new occupations that didn’t exist today. Did you see that little amusement of that, that AI story?
AI STORY:
Here are 10 common occupations in 2053: Lunar miner, bioprinter technician, climate engineer, cybersecurity analyst, virtual reality therapist, nanomedicine specialist, blockchain developer, augmented reality designer, gene editor, space tourism guide.
JAIME:
Yeah, I loved all the outer space stuff.
REID:
[laugh] Yes. Although by the way, when I read it, it was like, well, that was clearly an instance where it would’ve failed in an interview test for me, because it’s like, why would you send a human to do lunar mining versus a robot?
ARIA:
Yeah. [laugh]
REID:
Right? This is not actually thinking it through, right? [laugh] So, what did you think about those occupations, and which other ones do you think, just to help people think, of like, well here are new things that are going to be coming because of this. And obviously it’s always, you know, through the glass, darkly, in a complicated, very fast moving environment.
JAIME:
So, GPT-4 was a help here as well. I thought that the jobs were a little too predictable, so I asked for help coming up with, sort of, bigger changes, or jobs that looked more drastically different. And the suggestion was, you know, how much jobs are going to change, according to GPT-4, was really a function of three things.
One, the degree of complexity, uncertainty, or novelty involved in the job, task, or problem. And I think that the idea there is that, you know, we’re going to want those jobs that are really uncertain or new are places where people are going to be leaning in. Another was the degree of human empathy or emotion or ethics required for the job. And the last one was the degree of regulation, resistance, or risk associated with the job and its impact. So, looking at the jobs that were suggested, that would put things like virtual reality therapist, or gene editor, or even space tourism guide at sort of the high level of jobs that are going to be important in the future. And then, as exactly as you were saying, Reid, things like lunar minor or cybersecurity analyst down at the other end of potential jobs.
ARIA:
All right, I’m glad I’m not going to have to go to the moon. I’m not going to lie. [laugh]
SIMONE:
My name is Simone Stolzoff, and I’m the author of the book The Good Enough Job: Reclaiming Life from Work. The most interesting stat about how we work today is that 40 years ago, the average American and the average German worker worked the exact same number of hours each year. Today, the average American works 30% more than the average German. My hottest take is that by the year 2043, universal healthcare is going to become a reality in the United States.
Part of the reason why our relationship to work is so fraught is that the consequences of losing work are so dire. The United States spends a whopping 40% more per capita than any other country in the world on healthcare. In the next 20 years, our country will radically redesign the way we offer healthcare and decouple our basic human needs from our employment status. It’ll be one of the most revolutionary shifts in cultural opinion on the par of the federal legalization of gay marriage, or states legalizing cannabis.
ARIA:
So, another part of your research that we were excited about was your team did some great research putting EEG caps to monitor activity on people’s brains to see where people’s brains lit up, to see where people were stressed out. And I would just love to hear, like, what were some of those specific findings and how can we use them to make our current and future jobs, you know, even better?
JAIME:
Yeah, so we did a bunch of studies to try and understand the impact of remote and hybrid work on people. And we studied it in a number of ways, and EEG was one of the ways that we studied it, in part because one of the things that you can measure fairly well from brain studies is stress. So we ran these studies, and these are small scale, this is we bring people into the lab and are studying them. They’re looking at the stress people feel working in back to back meetings while remote. You would see, for example, that that process of going in back to back meetings increases your stress over the course of the day, but that if you do something like take five-minute breaks between the meetings or, even better, step outside and look at nature for a minute, that that significantly reduces the stress and then you can show up better at those meetings. And I think it makes a really important point as we’re thinking about the future of work, even in the context of AI, you know, we talk a lot about how we can start automating the drudgery or the repetitive parts of work. Sometimes those pieces of work are actually important for human attention. You know, I think about doing the dishes and that’s sort of meditative for me. [laugh]
I even think about how I start working on a PowerPoint document. I often like to sit and futz with, like, the bullets and the formatting before I get into the depth. And we’ve done a lot of research on studying the transition of attention, as well, and I think what those studies really suggest is how it’s not just about like, okay, raw, what do people do well, what do the large language models do well, how do we bring them together to make it better? But it’s like, how do we set people up to contribute the best, to think well, to see things in new ways. And that’s not always just raw, like, “okay, come up with a great idea, tell it to me now.” [laugh] And interestingly, one of the things with the brain studies that I think is going to be really fun, you can also, like, joy is another thing that you can see pretty well. And I just think of how much fun it is to work with these models and see things in new ways. And I think there’s a real opportunity to lean into that joy.
REID:
Do you think it’s going to be more, you know, like, you’re going to discover a certain number of people are going to be summarizing their documents into poems and sonnets as part of the joy of that? [laugh] I’d be curious, like, do you think your particular way of actually, in fact, using these tools to re-engage your own delight as a way of bringing mind and focus and attention. Do you think that’s going to be a broadly adopted thing? Or maybe not all poetry but a variety of them?
JAIME:
Yeah, absolutely. I mean, joy and magic has been really at the forefront of what we’re building. One of the things that was interesting that we learned from GitHub Copilot, for example, as we were going through it, there’s sort of all these standard metrics you can use to determine the quality of the suggestions you’re making to people as they’re programming. And you know, we like to say, “okay, we want to help people save as many keystrokes as possible.” So we put a lot of effort into like, okay, how often are people accepting the suggestions and how many characters have been saved? And it turns out actually the way to optimize that is to get quite short with the coding suggestions that you’re making, because then they’ll accept it and they won’t change it and they’ll move forward with it. But when we did that, even though our like metrics went way up, we got a lot of fuss from people who are like, “wait, where are those really magical, long suggestions that were showing up?” And we figured out that, like, that’s an important part that being able to see things holistically and think about that, that was important for people to get things done. And so like, really leaning into that magic and what does that mean and how are you seeing things differently is important.
REID:
Well one of the things we’re definitely going to have to do is, you know, kind of feed this transcript or some portion of it into GPT-4 and say, “give us a poem or sonnet summary or something!”
AI SUMMARY:
Reid, I can help with that:
In work there’s value in taking time,
but also in short, intentional bursts.
New perspectives can inspire, we find,
by changing places, problems seem reversed.
The rapid shift to working from afar
and AI’s growing role are intertwined.
Remote work breaks down boundaries, by far,
and leaves old definitions redefined.
Summarizing texts as poems, a new trick,
can bring a little joy to meetings dry.
With prompts to guide us, work’s finer and quick,
as we learn how to ask, “and what, and why?”
But work-life balance must remain in sight,
for wellness matters in the work we write.
ARIA:
I mean, I really do think that Reid has created more poems with GPT-4 than any poet has created previously. So, it really can bring out people’s creativity and magic and whimsy. Like, the future of work doesn’t just have to be, to your point, about like brute force data, like “ah, we improved it 1%!” It’s, “oh we improved people’s joy 10%, or we improved their, you know, enthusiasm 10%.” Like, those could be important numbers that we’re looking at.
JAIME:
Well, and this is so important too, right? One of the things that’s really interesting is what we want, as people, isn’t to do less. We want to do more meaningful stuff. And I actually was remembering, during COVID I needed to take a vacation so I took a week off, but you can’t travel anywhere, so I’m laying in bed and all I did for the first two days was play on my phone and watch Netflix. And I was miserable. Like, I was in tears, literally, and I was yelling at everybody, yelling at my kids, yelling at my husband. And then I decided to clean the house, like, I took one day of the vacation and I did one room at a time and just totally tore it apart and cleaned up. And it was so fulfilling and fun! [laugh] Like, it really meant a lot to me to be able to do that. And so, I think that that ability to, you know, it’s not about doing stuff for us, it’s about leaning into the meaningful work.
ARIA:
That really transitions well to my next question, because when we had Trevor Noah on the pod, his dream was to have 4-hour work days. And then, we’ve just seen the research coming out of England, they had that small pilot where they saw that 4-day work weeks increase productivity. And a lot of my questions would be about, you know, was that a short term that it increased or, you know, kept productivity at the same levels because it was novel and exciting? And what would that look like over time? And I would love to hear what you think, like, are these 4-day work week ideas, smart, not smart, too small scale? How would you think about that?
JAIME:
I actually think a 4-day work week is not the right thing to aspire towards, but like, getting stuff done and like sometimes that’s going to mean I’m working 24/7 cuz it’s super exciting and sometimes it means I’m at the dog park with my puppy all day. [laugh] And that flexibility, I mean, and that’s one of the things that the pandemic really, like, it created a ton of flexibility in how we get things done. And now we need to start figuring out how to use that flexibility properly.
ARIA:
Well, I think it also highlights that different people might need different things. Some people need flexibility because they have a dog, some people have kids, some people might have a disability that means that they need to work at different times of day. And so how can AI help us smooth that out? Because there are some problems with asynchronous work, you know, there’s benefits in us all, you know, being on the computer at the same time, but to your point, that flexibility can help us a lot. It’s not about working less, or only working 4 days, it’s figuring out what is right for you, Jaime, which might be different from Reid, or someone else.
JAIME:
Correct.
But if we weren’t all on the call right now, we wouldn’t be talking. So, you’re exactly right, the flexibility matters, but we’re social animals. We live, we collaborate, we work together and that matters as well. And that’s been one of the real challenges, is figuring out how to balance that flexibility. So, I think it’s really starts becoming about setting us up to succeed when we’re together. And AI can help us with that as well. We’re doing a lot of research into, for example, how to co-optimize schedules so that, you know, you can both maximize your own personal preference and your own flexibility and maximize our joint needs as well.
ARIA:
We’ve talked a lot about, again, this pandemic ushered in these questions, AI ushered in these questions. What’s an aspect about the future of work that people aren’t talking about that you think they should be talking about more?
JAIME:
Collaboration. I think we really should be thinking about collaboration more. And that, that was a big part of the pandemic, but a lot of remote work, a lot of what you do while you’re remote is actually your individual work. You’re very good at getting your own stuff done when you’re at home. Going into the office is about collaboration, it’s about unlocking other people, getting new ideas. And so, in some ways, that is a me versus we balance going on there. And, actually, I remember this, when I first went back to the office a couple of days a week, I’d be like, “oh my gosh, I got so much less done. I’ve got a pile of emails, I haven’t responded, I didn’t do anything,” and it turns out I did a lot. It just wasn’t the same kind of work, and I had to figure out that balance and transition again from like, oh yes, I have a super productive me, to also thinking about the we. And I think we’re likewise going to do that with AI. Like, how am I making a super productive me and how am I making a super productive we?
REID:
Well, one of the things that I think we already touched on some, but I want to come back to a little bit is, you know, most often people tend to think about work entirely on the outputs and efficiency. Which, of course, is an important part, but they don’t tend to talk as much about, like, the joy and the engagement, because, by the way those have the longitudinal, they have the endurance, they have, you know, the staying in a job, having those weeks that are 6-day weeks versus 5-day weeks or 7-day weeks, because you’re just like, it’s crunch time and you’re in it and you love it.
JAIME:
I’m just coming off a lot of 7-day weeks. [laugh]
REID:
Exactly. Well, thank you for spending your dog park time with us. [laugh]
What are some of the things that are, kind of, as people think about the architecture of work, what are going to be some of the angles to think about these broader variables? What do you think are some of the other variables that are going to be important for us to think about, you know, work design and team design and, you know, that kind of stuff?
JAIME:
Good, that’s a great question, and actually you’re making me think very much of measurement and how we understand what people are doing and you sort of started that out that way. We, at the moment, have very naive measures of productivity. You know, the best ways we measure productivity will be like, oh yes, number of keystrokes saved or the amount of time saved or maybe the number of emails you sent, which is certainly not something we want people to be, like, trying to increase or optimize. [laugh] What really matters is getting useful stuff done and our ability to get richer, better measures of what we’re doing. And this relates directly to the goal-directed AI, too, like translating from easy to measure numeric outcomes to meaningful outcomes, and large language models are amazing for that. And it actually makes me think of some data that we found in a recent work trend index that we have, where we found that people think they’re being very productive – this is particularly during the pandemic – people are super productive and pay a lot of attention to the remote work, you know, to the work that they’re doing, whereas employers actually are concerned about that. And, you know, I think it was 87% of employees reported that they were being productive at work, but on the opposite end, 85% of leaders said that the shift to remote work was making them question whether people were being productive. And I think that partly has to do with the wrong measures, and it may even be like, “oh, I’m being very productive, I’m stressed, I don’t know what’s going on, I’m doing lots of work, so I’m working hard, but I’m not producing the important outcomes.” And as we increasingly are able to understand and measure those outcomes, we’re going to be able to be more productive and get things that matter done.
RYAN:
It’s Ryan Roslansky from LinkedIn. For me, the most significant fact about today’s work environment relates to the rapid pace of change. LinkedIn data shows that, on average, if you look at the same job listing in 2015 versus 2023, 25% of the skills needed to do that job have changed. And it’s apparent that AI will accelerate this even further. So even if you aren’t changing your job, your job is changing on you. Thus success for individuals and companies over the next decade hinges on adapting to technological advancements like AI and embracing a skills-first mindset. By understanding our existing skills and identifying those future opportunities, we can navigate these changes effectively and better thrive in an ever-shifting economy.
ARIA:
The pandemic opened up so many types of work that some of them were positive. We realized how many types of work were knowledge work, we realized how many things could be done remotely. We gave people flexibility, Like, there were some silver linings that came out of this global pandemic, but we also saw so much burnout.
And it seems as if that burnout has stayed with us. I think it’s like an all-time high, 42% of workers globally report feeling burnt out. Like, why hasn’t that ebbed, what could we do about that? Do you see anything about preventing burnout or sort of returning to where we were before?
JAIME:
Yeah, well things have been changing so much. I’m like, didn’t we just deal with a crisis in work with the move to remote work? And now we’re dealing with a major reimagining with AI? Like, it is tiring. Personally, the thing that I found was leaning into the work and the opportunity. When I think about all that’s been happening, it wears me out. When I think about the opportunity with AI, I’m, like, so excited. We started building all of this cool AI, and all of a sudden I’m working harder than ever and really excited about what I’m doing.
There is real challenge and we have to figure out how to deal with all the disruption. Something like 53% of employees care about their health and wellbeing at a higher level than they did before. Like, that’s a strong signal, just like the data, there’s a strong signal that we need to be helping people figure things out. We’re not our best selves when we’re burned out, we’re not our best selves when we’re stressed. It is when we are relaxed, when we are safe, when we understand the world, that we’re able to do great things.
REID:
And one of the things I think we’re going to see through these various chat bots is helping on those variables too. It isn’t just going to be the fill in the forms or, you know, help write the memo – all super useful – or take the meeting notes, or remember the action items, or get the information to so-and-so. I think this is all in the thread of, not just the productivity, it’s actually in the human engagement in the delight and the fun. I think the: summarize this as a poem, for example, make it as a rap song, or put this in iambic pentameter, and it can be like The Odyssey or The Iliad.
One of the things that I’m doing as part of the – and you’ll get this, too – as part of the personalized version of the Impromptu book are, like, epic poems that are kind of like, you know you, me, and AI, as a way of doing it.
So, let’s do a few of the rapid fire questions. Is there a movie, song, or book that fills you with optimism for the future?
JAIME:
So, I reread and rewatch things over and over again, so my answer here is going to be somewhat trite, but something that I’ve seen a lot is probably Star Trek: The Next Generation. [laugh] And I just restarted it again, I’m on episode five now. It was something I watched early in the pandemic and that I’m watching now for slightly different reasons, sort of helping me process things differently. During the pandemic, it was really cool to have this, like, you have this small little world with all the same characters, so it’s very much – and they’re always playing concerts and putting on plays and it’s made me think of being stuck at home with my 4 boys and my husband, but then you’re traveling the galaxy and seeing all of these other places. So I thought it sort of fed my pandemic need.
And then now, with AI, it’s really interesting in the obvious ways. You know, you’ve got data and you’ve got the computer and it’s funny to be watching it again now where when I was watching it not that long ago, I’m like, “oh my gosh, this is…” Oh yes, Star Trek actually has all sorts of interesting human/computer interaction affordances that they are bringing, but the computer stuff is all made up. To really be rewatching it and be like, “oh, and we could do better than that, even!” [laugh]
REID:
When you get your Impromptu book, a Star Trek episode is also a prompt of, yes, a prompt in there. One of the things I did is when I was an undergraduate at Stanford is I actually taught one particular, like one class, not a quarter class, but one class around one of the Star Trek episodes, “The Measure of a Man,” which was the one with data on, you know, whether he had to follow an order that was about dismantling himself. And I thought it was a great encapsulation of a whole set of different issues. So I literally, you know, the asynchronous work, right, was go watch it and then we’re going to pose some questions and talk about it.
ARIA:
Awesome, I love it. All right, so question 2. Obviously, future of work, AI super top of mind for you, but where do you see progress, momentum outside of your industry that inspires you for the future?
JAIME:
So, nothing is outside of that right now. [laugh]
ARIA:
True, thanks. [laugh]
JAIME:
That being said, I’m really excited about education right now and what that means. And, you know, it’s because the questions that we’ve been talking about this whole time, they’re the fundamental questions of like, what does it mean to exist in the world? What does it mean to make meaningful contributions? And I think that’s really interesting. And certainly education was deeply impacted by remote work and it’s going to be deeply impacted by AI as well, and I think we’re going to need to do a lot of deep thinking as well about these new skills that we want the next generation to learn and ourselves to learn.
REID:
Obviously you’re paying a lot of attention to technology and, like, AI, and impact on the work, and impact on research, and impact on all the rest. Is there any technology, and maybe it’s also AI is the answer here, but is there any technology you’re watching to make sure it stays on course, to make sure that, you know, the guardrails are there and so forth? And maybe AI is the answer, and in which case, what are the, in addition to the amplification, what are the guardrails you’re paying attention to?
JAIME:
I think the thing that I’m particularly excited about and anxious about and paying a lot of attention to right now marries two deep interests of mine. So, I have a PhD from the MIT AI lab, but I actually got it studying information retrieval. And right now, there’s this really interesting opportunity that we talked about a little bit, to, like, take these reasoning engines and couple them with grounding knowledge and building on that. And that’s really leaning into the language models where they’re different from us. That ability to go and search, like, search engines are amazing and the ability to index all of the world’s knowledge and all of your emails and all of the information you have access to and, like, as we start capturing more index, and surface that and use it, it’s really interesting. And so, I’m excited about that.
ARIA:
Well, we’d love for you to leave us with one final thought. So, we always say, you know, what is possible to achieve if everything breaks humanity’s way? So, in the next 15 years, where do you think we can be if everything breaks our way, and what’s the first step? How do we get there?
JAIME:
Oh, well I hope to be exploring the universe Star Trek-style, although it doesn’t need to – it can just be on this planet. And I think the first step is really leaning into what makes people think well.
ARIA:
Keep that curiosity up and we’ll be okay.
REID:
Well I think those are amongst the index of skills, you know, ranging from empathy to all the other ones we were talking about, collaboration. Keeping your curiosity up is going to be one of the skills that our AI copilots hopefully are going to help us with.
JAIME:
Yeah, for sure. So far it is.
REID:
Jaime, as always, thank you and thank you for after many 7 day weeks, coming on to talk with us. [laugh]
JAIME:
It was my pleasure.
ARIA:
So, what I loved about talking to Jaime on this auspicious day in particular, the day that, you know, Microsoft launched M365 and is literally showcasing how AI can be a copilot for your job, for your profession, for making things better, is she is a creator but also a consumer. I mean, throughout the pod she was like, “oh, you asked me that question, I popped it into GPT-4. This is what they said.” You know, she talks about how every email or every document she’s getting, she’s getting bullets, she’s getting a summary. She also mentioned she’s getting it in poem form, and I love that she was using her own products to actually enhance her own productivity, but also didn’t forget about the whimsy and the fun and the joy, which is often forgotten. So, I loved that she mentioned that as well.
REID:
Typically, when you go to like a research scientist who’s been researching work, you’re thinking you’re going to get, “productivity x,” or, “productivity y,” or, like, you know, “you find that intervals of 8 minutes sprints followed by, you know, 1 minute rest breaks.” But the fact that it was like, “well, actually, in fact, here’s how I keep myself creatively engaged and here’s how I bring the light to it, and we need metrics for those things, too. And the way I do it is, yes, we’ve just launched all these really cool copilot summary features.” Those things are not just for important imaginations in the future work, our imagination in the use of GPT-4 and tools, but also the things to do both as individual workers but also as designers and creators of the future. And I think that was part of the unexpected delight of this interview.
ARIA:
The two of you also spoke about, which I thought was so good, moving from the I and the me, like, my work is going to be enhanced, my work is going to be better. Well, what about our work? What about the work we’re doing collectively? I just remember when you’re, you know, you’re a junior in high school and you get assigned a group project, you’re like, “ugh, this is the worst. I have to work with them, like, I’m so much better.” And you don’t quite realize that no, you’re not good unless you can work on a team. And so that’s just so critical to the future of work and how AI can enhance that teamwork, not just you as a solo contributor. And that’s so critical for what we’re doing.
REID:
Yeah, it was one of the things that reminded me of my first book, The Startup of You, because it’s like, life is a team sport, not an individual sport. It’s actually one of the things I think the educational system, through its, like, “how do you measure individuals,” and so forth, gets fundamentally wrong. But actually, in fact, every class should have group work components because almost all the work, even writing a novel, is a group work thing. [laugh] Right? So, that group work and the fact, look at the tools not just as individuals, like, you know, “I am here with my trusted copilot sidekick, so we are here with our copilots!” And it’s the we about how we’re working that is, I think, such a valuable lens into the future work.
ARIA:
Yeah, and it’s like, how can AI be a trusted team member and copilot for what we’re doing, but, again, for the whole team? And maybe you have a different LLM for what you’re doing personally, you have a different one for your group. You have a different one for the group over there. And so, again, just remembering that I think is really great.
REID:
Possible is produced by Wonder Media Network, hosted by me, Reid Hoffman and Aria Finger. Our showrunner is Shaun Young. Possible is produced by Edie Allard and Sara Schleede. Jenny Kaplan is our executive producer and editor.
ARIA:
Special thanks to Caitlin Cummings, Katrina Zuccaro, Lauren Cole, Surya Yalamanchili, Saida Sapieva, Ian Alas, Greg Beato, and Ben Relles. And huge gratitude to Papia Debroy, Simone Stolzoff, Ryan Roslansky, Jessi Hempel, and everyone who called in with their thoughts on the future of work. Thanks so much.