The Future of Work: Do Humans still matter?

00:00:00: ,260 [Andrea] Will AI take your job?

00:00:18: ,370 [Andrea] Or even more radical, will humans even matter economically anymore?

00:00:25: ,940 [Andrea] Welcome to a new episode of the Sharp 10 Global Trends podcast.

00:00:30: ,400 [Andrea] My name is Jessica, and today I am reading the observations and questions of Andrea. He believes my voice is clearer and easier on your ears. And despite him being a man, he had a strong wish that a woman's voice leading this conversation feels like the right kind of diversity for a podcast about the future. I hope you agree. It's March twenty twenty-six, and the headlines are getting tough to ignore. UPS just laid off forty-eight thousand people blaming their new AI driven network of the future. Amazon has already cut another sixteen thousand this year on top of fourteen thousand last year. Accenture, Citigroup, Goldman, Klarna, Lufthansa, one after another, companies keep saying the same thing, AI made us do it. So what if humans just are not or less needed in the economy anymore?

00:01:25: ,040 [Andrea] Who actually owns this new AI economy? And if work stops defining us as humans, what replaces it?

00:01:34: ,870 [Andrea] To dig into that, I am bringing in one of my favorites in the AI world, Claude, the large language model from Anthropic. Claude's replies and explanations are read by Charles.

00:01:47: ,740 [Andrea] On March fifth, Anthropic published a remarkable blog post called Labor Market Impacts of AI, A New Measure and Early Evidence. Perfect timing for this conversation.

00:02:00: ,980 [Andrea] All right, let's get into it.

00:02:04: ,440 [Andrea] drum roll] Claude, what is your take on the mass layoffs at the companies I just mentioned?

00:02:11: ,640 [Claude] It's a pleasure talking to you again, Andrea. The layoffs are real and significant, but I think we need to be careful not to confuse two things happening at the same time. On one hand, yes, AI is automating tasks that humans used to do. Klarna is the clearest example. They have been remarkably candid that their AI assistant is handling work that previously took hundreds of customer service agents. That is not spin. That is a measurable operational shift.

00:02:44: ,100 [Claude] But many of these layoffs are also happening for reasons that predate AI.

00:02:50: ,740 [Claude] UPS has been restructuring its delivery network for years. Goldman Sachs has been trimming headcount through multiple cycles since twenty twenty-two. When a CFO stands up and says, "AI made us do it," there is sometimes a convenient narrative at work because AI is a socially acceptable explanation in a way that we overhired during the pandemic boom simply is not.

00:03:18: ,100 [Claude] So my take is this, AI driven displacement is real, it is accelerating, and it will get much more significant over the next three to five years. But right now, in early twenty twenty-six, we are still in a phase where AI is amplifying human productivity more than it is replacing humans wholesale. The Anthropic research you mentioned actually supports that nuance. Their data suggests AI exposure correlates with wage growth in many sectors, not just job loss, which leads to a more uncomfortable question than, will AI take your job? It is, whose productivity gains does AI actually benefit?

00:04:02: ,040 [Andrea] Can you explain that a bit further?

00:04:05: ,700 [Claude] When we talk about AI amplifying productivity, what that means in practice is that one person can now do the work that previously required three, five, or ten people.

00:04:16: ,320 [Claude] A single lawyer with the right AI tools can review contracts at a speed that used to require a whole team of junior associates. A marketing manager can produce campaign content that used to need an agency.

00:04:31: ,640 [Claude] A software engineer can ship features in days that used to take weeks with a larger team.

00:04:38: ,900 [Claude] Now, the critical question is, when that productivity gain happens, where does the value go?

00:04:45: ,400 [Claude] In most cases right now, it flows to the company, specifically to shareholders and to some degree, senior leadership.

00:04:54: ,200 [Claude] The lawyer's firm does not hire two more lawyers because one lawyer is now three times as productive. It lays off two lawyers or simply does not hire replacements when people leave and the profit margin improves.

00:05:09: ,900 [Claude] This is what the Anthropic research is pointing at when it finds wage growth in AI exposed sectors. The people who keep their jobs in AI adjacent fields, the ones who learn to work with these tools effectively are becoming more valuable and being compensated for it.

00:05:28: ,560 [Claude] A data scientist, a strategist who knows how to collaborate with AI rather than compete against it, these people are seeing their market value go up.

00:05:38: ,660 [Claude] But that creates a bifurcation. At the top, a relatively small group of highly skilled workers whose productivity and wages are rising sharply,

00:05:49: ,220 [Claude] and at the bottom, a much larger group whose tasks are being automated away and who do not yet have the skills to move into that upper tier.

00:06:00: ,514 [Claude] So the productivity gains from AI are very real. The economy is becoming more efficient.But more efficient does not automatically mean more equitable. And that gap between who generates the value and who captures it is arguably the central economic tension of this entire AI transition.

00:06:23: ,794 [Andrea] So if I understand you correctly, AI is not just taking jobs, it is transforming them. It is replacing people who do not know how to use it, but boosting the ones who do. Is that what you are saying?

00:06:37: ,724 [Claude] Yes, but with one important caveat. It is not just about knowing how to use AI. It is about whether your core judgment, creativity, or relationships are still needed after AI handles the execution, because AI can now do a lot of the execution.

00:06:57: ,494 [Claude] The people who are most protected are not necessarily the most technically sophisticated. They are the ones whose value was never really about task execution in the first place. It was about deciding which tasks matter, building trust with clients, reading a room, making a call under uncertainty.

00:07:19: ,564 [Claude] A mid-level analyst who spent their career running reports, vulnerable.

00:07:25: ,604 [Claude] A senior advisor who spent their career being right about hard calls in ambiguous situations, much less so because AI actually makes them more powerful by removing the grunt work.

00:07:38: ,054 [Claude] So the real dividing line is not uses AI versus does not use AI.

00:07:45: ,664 [Claude] It is whether your value is fundamentally about doing or about thinking, judging, and connecting. That is the uncomfortable part for a lot of executives right now, because many middle management roles were always more about coordinating and reporting than judgment, and AI is very good at coordinating and reporting.

00:08:06: ,943 [Andrea] But let me push back on the pessimistic framing for a moment, because there is a historical argument that I find compelling, and I do not think we have fully addressed it yet.

00:08:19: ,524 [Andrea] If you look at the Industrial Revolution, automation did not lead to mass permanent unemployment. Quite the opposite. Yes, it displaced weavers and blacksmiths, but it also created entirely new industries, new categories of work, and a level of prosperity that nobody in 1820 could have imagined. More people ended up working, not fewer. Countries and companies became dramatically more productive. Take a simple example. That law firm you mentioned earlier does not have to shrink because of AI. It could choose to grow. It could keep its lawyers, even bring in new ones, take on more cases, or dedicate time to pro bono work and truly build a name that stands for something.

00:09:05: ,444 [Andrea] And it is not just law firms. Every company has this opportunity right now, not to reduce people, but to unlock their potential, to raise the bar in customer service, to rethink products from the ground up, not just responding to what customers ask for, but creating experiences so good, so thoughtful that they redefine expectations entirely. And maybe even more powerful is this. Let people create, let them invent, let them use what they see, what they learn, what they experience every day, and turn it into something new. Because every major technological shift in history has done more than replace jobs. It has opened doors. It has created roles that did not exist before. It has given people new ways to contribute, to build, to matter. Just think about what AI will help discover or invent that we can't even imagine yet.

00:10:08: ,964 [Andrea] So yes, some things will disappear. That is part of progress. But from that change, something new can rise, new roles, new ideas, new opportunities, and many of them will need people, curious and creative people, people who are ready to shape what comes next.

00:10:28: ,153 [Andrea] So why should this time be fundamentally different to any technical revolution before?

00:10:34: ,224 [Claude] It should not be automatically assumed to be different, and the historical analogy is important. It is the strongest argument against the apocalyptic reading of AI and work. Every major wave of automation in history has ultimately expanded the economy rather than contracted it. The cotton gin did not eliminate textile workers. It expanded the textile industry so dramatically that employment in it grew. The same pattern played out with electricity, with computing, with the internet.

00:11:08: ,024 [Claude] The question is not whether AI will follow that pattern. It probably will eventually. The question is the speed of the transition and whether we are building the conditions that make the optimistic outcome actually happen rather than just assuming it will.

00:11:26: ,824 [Andrea] And here is what I think makes this moment potentially different from previous transitions in a positive sense. Not just that AI handles execution while humans do other things, but something more specific. What if companies deliberately invest in developing the human capacities that AI cannot replace? The thinking, the judging, the connecting. What if instead of just cutting headcount, organizations use AI as an engine room underneath a workforce that is now operating at a higher cognitive level across the board? Not just a few senior people thinking strategically while everyone else executes, but an entire organization where even entry-level workers are freed from execution and push toward judgment, relationships, creativity.Because AI handles the rest

00:12:19: ,954 [Claude] That is the exciting scenario, and it has a name in the academic literature. A new MIT paper published just last month, Some Simple Economics of AGI by Christian Catalini and colleagues, calls it the augmented economy.

00:12:36: ,464 [Claude] The alternative, which they model with equal rigor, is what they call the hollow economy, where AI executes at scale, but nobody is actually verifying, thinking, or taking responsibility for the outputs. Measured activity rises, but hidden risk accumulates underneath. The economy looks productive on paper while silently building up misaligned, unverified output.

00:13:01: ,584 [Claude] The difference between those two futures is not technological. The technology is the same in both scenarios. The difference is organizational and cultural. Do companies invest in human judgment and verification as AI scales,

00:13:18: ,884 [Claude] or do they just cut costs and hope the outputs are good enough?

00:13:22: ,344 [Andrea] Which maps directly onto the Industrial Revolution analogy. The countries and companies that navigated that transition well were not the ones that simply deployed the new machinery.

00:13:35: ,824 [Andrea] They were the ones that also invested in education, in workforce development, in the institutions that allowed new capabilities to emerge. The ones that just automated without investing in people ended up with social rupture.

00:13:51: ,124 [Claude] Exactly. And the MIT paper adds a sharp warning here that is worth taking seriously. They identify what they call the codifier's curse.

00:14:02: ,744 [Claude] The idea is this: the experts who are best placed to develop and transmit the next generation of human judgment, the senior lawyers, analysts, strategists, designers, are often the same people being asked to train AI systems to do their jobs.

00:14:19: ,704 [Claude] Every time they do that well, they accelerate their own obsolescence. The better you are at codifying your expertise into an AI, the faster you make yourself replaceable. So there is a real tension at the heart of your optimistic scenario. The people who should be building the human capabilities of the next generation are simultaneously being incentivized to transfer those capabilities to machines. That is not a reason to abandon the optimistic vision, but it is a reason to be deliberate about it rather than assuming it will happen automatically.

00:14:56: ,444 [Andrea] So verification itself becomes a job. Is that what you are saying?

00:15:01: ,584 [Claude] Yes, and I think it is one of the most underappreciated ideas in this entire debate.

00:15:08: ,264 [Claude] In the current economy, the scarce resource is intelligence and execution. AI is making both of those abundant and cheap, so the value migrates, as it always does in economics, to whatever becomes scarce next. And according to the MIT paper, that is verification, the capacity to look at what an AI has produced and evaluate whether it is right, whether it serves the actual intent, whether it is safe to act on.

00:15:39: ,234 [Claude] But verification is not just spell-checking an AI output. It is much deeper than that. It includes domain verification. Does this legal brief actually reflect sound legal strategy, not just plausible-sounding language?

00:15:56: ,484 [Claude] Intent verification, does this output serve what the client actually wanted, not just what was literally typed into a prompt?

00:16:05: ,784 [Claude] Risk verification, what could go wrong here that the AI did not flag?

00:16:11: ,224 [Claude] And ethical verification, is this aligned with values and responsibilities that were never written down anywhere?

00:16:20: ,344 [Claude] None of that can be done by someone without expertise and judgment, and the gap between what AI can execute and what humans can afford to verify is growing faster than the supply of qualified verifiers, which means verification becomes increasingly scarce and increasingly valuable.

00:16:40: ,993 [Andrea] Which completely reframes the career ladder. You are not promoted because you executed better than anyone else. You are promoted because you developed the expertise to catch what AI gets wrong.

00:16:54: ,124 [Claude] Precisely, and that requires a very different kind of professional development than we currently invest in. The traditional path from junior executor to senior strategist worked because execution and judgment were bundled together. You learned judgment by doing execution, making mistakes, and gradually developing deeper expertise. The MIT paper calls the collapse of that pathway the missing junior loop. When AI handles the execution, the apprenticeship model breaks down, and you have to find a new way to build the expertise that makes verification possible. There is also a painful irony here. At exactly the moment when verification becomes most valuable, the pipeline that creates expert verifiers is under threat. Fewer people are developing deep domain expertise through hands-on work precisely because AI is doing that work instead. So the organizations and societies that invest deliberately in building verification capacity right now through education, through mentorship, through creating roles where humans develop expertise rather than just supervise outputs, are going to have a structural advantage over the ones that do not.

00:18:12: ,684 [Andrea] And the ones that do not?

00:18:14: ,540 [Claude] They risk building the hollow economy.Not a dramatic collapse, but a quiet accumulation of unverified, misaligned output underneath a surface of impressive-looking productivity metrics.

00:18:28: ,190 [Claude] Every organization deploying AI without investing in human verification is quietly building that risk on its balance sheet. It just does not show up until something goes badly wrong. A legal brief that looked correct but was not. A financial model that passed every automated check but violated the actual intent. A medical recommendation that satisfied the measurable criteria but missed what the patient actually needed. The MIT paper offers one practical framework for companies to avoid this. They call it the AI sandwich. Human intent at the top, someone defining what actually needs to be achieved and why. Machine execution in the middle. AI handling the speed, scale, and processing.

00:19:20: ,400 [Claude] Human verification at the bottom. Someone with expertise checking that the output actually delivers on the intent. That three-layer structure is simple, but building it properly requires investing in the human layers, not just the machine layer. Most companies right now are investing heavily in the middle and quietly hollowing out the top and bottom. That is the road to the hollow economy. The road to the augmented economy runs in the opposite direction.

00:19:50: ,140 [Andrea] Okay, so it is not as black and white as being replaced or not replaced.

00:19:56: ,840 [Andrea] It really depends on two things. The kind of work you do and how your company chooses to see AI. As a tool for growth or simply as a way to replace people, especially those focused on creative and strategic thinking and those who verify that what AI produces is actually right, meaningful, and valuable. But as of today, most people in the workforce are doing execution work, right? Roughly what percentage of people are actually executing versus being creative or strategic?

00:20:30: ,520 [Claude] The numbers are sobering. Depending on how you define it, somewhere between sixty and eighty percent of the global workforce is primarily doing execution type work. Routine cognitive tasks, process driven work, coordination, data handling, customer service.

00:20:49: ,600 [Claude] The International Labor Organization and various McKinsey studies have put the highly automatable share of global tasks in that ballpark. Now, not all of that disappears overnight, but a significant chunk is squarely in AI's crosshairs over the next decade. If you run a company of a thousand people, statistically maybe one hundred and fifty to two hundred of them are doing work that is hard to automate. The judgment calls, the relationship management, the creative direction.

00:21:23: ,040 [Claude] The other eight hundred are doing work where AI will, within five to ten years, be able to handle a meaningful portion of the execution. That does not mean eight hundred people lose their jobs tomorrow, but it does mean that every leader needs to be asking right now, what is the actual human value add in each of these roles?

00:21:46: ,060 [Claude] And if the honest answer is mostly execution, that is a role that needs to be redesigned, not just defended.

00:21:54: ,530 [Andrea] Let's get straightforward. What percentage of jobs are actually at risk of disappearing as AI agents become more capable?

00:22:02: ,880 [Claude] The most cited figures come from McKinsey, Goldman Sachs, and the Oxford Martin School. They converge somewhere around thirty to forty percent of current job tasks being automatable within the next ten years. But tasks are not the same as jobs. Most jobs are a bundle of tasks, some automatable, some not.

00:22:24: ,080 [Claude] When you translate that into actual job disappearance, the more likely range is probably fifteen to twenty-five percent of current jobs being either eliminated or reduced to a fraction of their current headcount over the next decade. That is roughly five hundred to eight hundred million jobs globally. The number I find most striking, Goldman Sachs estimated in twenty twenty-three that two-thirds of jobs in the US and Europe are exposed to some degree of AI automation. Not eliminated, exposed. Meaning they will change significantly and the people in them will need to adapt.

00:23:07: ,880 [Claude] And what catches people off guard is the speed. Previous technological transitions played out over generations. This one is playing out over a few years only. That is what makes it categorically different.

00:23:22: ,800 [Andrea] I actually came across something this morning that puts a much more concrete face on those numbers. Andrej Karpathy, the OpenAI co-founder, published an impressive interactive visualization of the US job market, scoring three hundred forty-two job types in AI exposure across one hundred forty-three million workers.

00:23:45: ,340 [Andrea] He was careful to call it a development tool, not a study or economic forecast. It went viral almost instantly and was immediately misread as a doomsday prediction, which is not what it says at all. But even as a rough map, it is illuminating. Around forty-two percent of American workers sit in high exposure jobs, and analysts applying the same methodology to other labor markets are finding consistent patterns with some variation depending on how manufacturing heavy an economy is.

00:24:19: ,188 [Claude] That caveat from Karpathy is worth holding onto throughout this conversation. These scores measure exposure to change, not probability of disappearing. A high score means the job is being transformed, not necessarily eliminated. The difference between those two things matters enormously for how people should be responding. Panic is not the right reaction, but neither is complacency.

00:24:46: ,288 [Andrea] And it confirms something I have been thinking about for a while. My instinct was actually more radical. If AI can work on a computer, then eventually every job done on a computer is potentially at risk. The whole white-collar screen-based world. What the Karpathy data does is put a more granular face on that. Corporate management and office administration, the email, the scheduling, the documentation, the coordination, the reporting scores around seven out of ten. Finance and accounting in the same zone, six to seven. And at the very top, software developers, graphic designers, data analysts, paralegals scoring eight to nine, which if you follow my logic, should not surprise anyone. These are almost entirely computer-based jobs.

00:25:39: ,548 [Claude] Your instinct could be right. If the work product is fundamentally digital, AI is coming for it. The question is not whether, but at what pace and what happens to demand on the other side. Software developers score nine out of ten because almost everything they do is digital, and AI is reshaping it fast. But a high exposure score is not a death sentence. Demand for software is accelerating, and a developer with AI tools is far more productive, not redundant.

00:26:14: ,008 [Andrea] And then at the other end, scoring one or two. Roofers, landscapers, electricians, plumbers, hands-on healthcare workers currently shielded. Though I want to add an important thought here. That shield only holds as long as we are talking about software AI alone. The moment you integrate AI with robotics, drones, and autonomous systems, that protection starts to look much more temporary, and that integration is already happening. Humanoid robots are moving out of the lab. Agricultural drones are already operational at scale.

00:26:54: ,008 [Andrea] The Karpathy tool explicitly measures digital AI exposure, not robotic exposure.

00:27:01: ,778 [Andrea] So the one and two scores are not a permanent safe zone. They are more like a delay.

00:27:07: ,658 [Claude] That is an important disclaimer. The physical work is safe narrative has a hidden assumption baked into it, that AI stays confined to a screen.

00:27:18: ,508 [Claude] But embodied AI is a separate wave of disruption running roughly a decade behind the software wave.

00:27:26: ,468 [Claude] The reason it has not hit yet is not that it is impossible. It is that hardware is harder and more expensive to develop than software. But the trajectory is steep. Warehouse automation is already displacing manual workers. Humanoid robots from companies like Figure, Unitary, Tesla, and Boston Dynamics are in early commercial deployment, targeting exactly the repetitive physical tasks that currently score lowest on AI exposure. So the more accurate picture is not physical work is safe. It is physical work has a longer runway, which is still meaningful, but it is a very different message.

00:28:09: ,428 [Claude] As AI compresses the white-collar labor market, more qualified people will compete for fewer screen-based roles while the physical trades remain chronically understaffed, at least until the robotics wave arrives and reshuffles the picture all over again.

00:28:26: ,868 [Andrea] So if that is the landscape for the workforce as a whole, let me bring it down to something much more personal. I have three young daughters, and I find myself wondering, what kind of world are they actually stepping into? What does all of this mean for someone just starting out?

00:28:44: ,408 [Claude] This is one of the questions that concerns me most about the AI transition because the entry-level problem is structural, not just cyclical. Historically, junior roles served a dual purpose. Yes, they got work done, but they were also how organizations transferred knowledge to the next generation. You learned by doing the analysis, drafting the memo, sitting in on the client call, making small mistakes in low-stakes situations. That apprenticeship model, even when it was not called that, was how careers were built.

00:29:20: ,088 [Claude] AI is now doing a lot of that entry-level work, and companies are discovering they can get the output without hiring the junior, which means the ladder is losing its bottom rungs.

00:29:33: ,848 [Claude] For young people entering the workforce right now, I think this creates three distinct paths. The first is the technical path, becoming an expert at building, training, or deploying AI systems. That is a real on-ramp, but it is narrow. Not everyone is a software engineer.

00:29:54: ,008 [Claude] The second is the human-intensive path, deliberately choosing fields where presence, empathy, and physical skill matter. Healthcare, education, skilled trades, social work. These are undersupplied, often undervalued, but durable.

00:30:12: ,036 [Claude] The third, and the one I would encourage most, is what I would call the judgment apprenticeship, finding organizations, mentors, or roles where you are explicitly being trained to think and decide, not just execute.Even if the execution around you is increasingly automated, the opportunity to develop judgment in real contexts is still available. But you have to seek it out deliberately. It will not come automatically from a job title anymore. The risk is that young people land in roles where AI does the interesting parts and they are left supervising outputs they do not fully understand, gaining neither skills nor judgment. That is the new dead-end job, and it looks fine on paper until it's not.

00:31:01: ,096 [Andrea] So if we follow that logic, we end up with three broad paths. Go deep into tech, move into roles that require real human presence, or find your way into positions that build creative and strategic thinking.

00:31:17: ,056 [Andrea] But here is an important question. That human intensive path, healthcare, education, skilled trades, social work, these are essential roles, and yet historically they have been underpaid. A nurse on a night shift often earns far less than a banker sitting at his desk trading stocks during the day. So as AI starts to take over more execution driven work, do we finally see a shift in how society values these deeply human roles, or does the imbalance stay?

00:31:49: ,616 [Claude] I think the imbalance will stay, but the pressure to change it is building in a way we have not seen before. Here is why it persists. Pay in most economies is not primarily determined by social value or even scarcity of skill. It is determined by proximity to capital and by how easily your output can be measured and monetized. A stock trader sits inside a machine that directly generates revenue. A nurse sits inside a cost center. That is brutal, but it is the structural reality that has kept care work undervalued for decades, regardless of how essential it is. But here is what is shifting. As AI systematically hollows out the white collar middle, the analysts, the paralegals, the junior consultants, the relative scarcity of genuinely human work increases, and scarcity eventually moves price. If you cannot automate the night shift nurse and you cannot offshore her, and demographic aging means you need three times as many of her over the next twenty years, at some point the labor market has to respond. We are already seeing early signals. Skilled trades in Germany and Switzerland are commanding wages that rival many university graduate roles. In the US, starting wages in healthcare have moved significantly post pandemic and have not come back down. But market forces alone probably will not be enough because care work has another structural problem. It is largely publicly funded and governments are slow, politically constrained, and often more focused on cutting costs than repricing essential labor. So my answer is the shift will happen, but unevenly and too slowly for the people living through it right now. The market will eventually price human presence correctly. Policy may or may not catch up, and in the meantime, a lot of nurses and teachers will keep carrying enormous social value on wages that do not reflect it.

00:34:02: ,526 [Andrea] Maybe the broader question is this, why do we actually need to work at all? Or taking a slightly less futuristic angle, could we start working significantly less as more and more execution is handled by AI agents?

00:34:17: ,036 [Claude] This is where the conversation gets philosophically interesting, and I want to separate two things that often get confused. The first is the economic question, do we need to work as much to sustain material wellbeing? And the answer is increasingly no. If AI agents can handle growing portions of execution, the productive capacity of society does not require the same volume of human labor hours it once did. That is not speculation. That is arithmetic. The second is the existential question, do humans need to work for other reasons?

00:34:57: ,552 [Claude] And here the answer is much more complicated because work for most people was never just about income. It was about getting up in the morning with somewhere to be. It was about being useful, about a community of colleagues who knew your name. Strip that away, even if you replace the income, and you have a serious human problem. We actually have data on this. Lottery winners report surprisingly low long-term happiness gains. Early retirement correlates with cognitive decline. Regions hit by sudden de-industrialization, whether in the American Rust Belt or parts of the UK after the mines closed, did not just experience poverty. They experienced a collapse in mental health, life expectancy, and social cohesion. The income loss was bad. The loss of purpose and structure was arguably worse. So the four-day work week conversation, which is gaining real traction in Europe right now, is interesting precisely because it does not abolish work. It compresses and redistributes it. Pilot programs in Iceland, the UK, and Germany have shown broadly that productivity holds while wellbeing improves. That feels like the pragmatic near term answer. But the longer term question is more radical. If AI keeps accelerating, we may be heading toward a world where twenty hours of human work per week is sufficient to run a modern economy.And at that point, society needs a serious answer to the question: what do humans do with the other twenty hours? And how do we ensure those hours have dignity and meaning, not just Netflix? Because leisure without purpose is not freedom. Historically, for most people, it is actually quite destabilizing.

00:36:53: ,692 [Andrea] You mentioned the four-day workweek. If you think about it, that alone could already absorb a twenty percent reduction in jobs just by spreading work among more people.

00:37:04: ,312 [Andrea] And then there is the universal basic income, an idea that has been around for years. What is your take on these approaches? And beyond those, what other models do you see for rethinking how we structure work?

00:37:19: ,122 [Claude] Let me give you an honest assessment of each and then point to what I think is the most underrated idea in this space.

00:37:27: ,512 [Claude] The four-day week is the most politically viable near-term lever. The evidence is encouraging. Pilots show productivity holding, absenteeism dropping, retention improving. For knowledge work especially, it is a relatively low-risk experiment. But it is a redistribution tool, not a transformation tool. It spreads existing work more humanely. It does not answer what happens when the volume of work itself shrinks significantly.

00:37:58: ,632 [Claude] Universal basic income is the idea that will not go away, and for good reason. The core logic is sound. If AI generates enormous wealth with less human labor, you need a mechanism to distribute that wealth more broadly, or you get extreme concentration at the top.

00:38:17: ,532 [Claude] Pilot programs in Finland, Kenya, and Stockton, California, have shown that people do not stop working when given a basic income. They actually make better decisions, invest in education, start small businesses. But universal basic income has two serious unsolved problems.

00:38:37: ,472 [Claude] The first is fiscal. At meaningful scale, it is extraordinarily expensive.

00:38:44: ,552 [Claude] The second is the existential problem we just discussed. Income without structure or purpose does not automatically produce flourishing.

00:38:54: ,712 [Claude] Beyond those two, the idea I find most compelling and underrated is what some economists call care economy investment. Essentially, society deliberately deciding to fund and professionalize the work that markets chronically underprice: childcare, eldercare, mental health support, teaching, community building.

00:39:17: ,832 [Claude] These are labor-intensive by nature, resistant to automation, and in massive undersupply demographically.

00:39:26: ,372 [Claude] A political decision to pay for them properly through taxation of AI-generated productivity gains could simultaneously create meaningful employment, address social needs, and rebalance what society signals it actually values. Think of it as the inverse of the current model. Instead of letting AI displace workers and hoping the market eventually reprices human labor, you deliberately redirect the productivity dividend into the work only humans can do well. That is not utopian. It is a policy choice. The question is whether there is the political will to make it.

00:40:07: ,092 [Andrea] Okay, so if I understand that last idea correctly, it's about starting to pay care or social jobs better by taxing the additional profits gained from the productivity increase thanks to AI agents, right?

00:40:21: ,332 [Claude] Right. AI agents generate productivity gains. Those gains flow to the companies that deploy them, showing up as higher margins and profits. You then tax a portion of those profits and redirect that revenue into the jobs AI cannot do: care, education, social work, mental health support. This isn't a radical idea. It's actually quite conservative in structure. Societies have always taxed productive activity to fund collective needs. This is just updating the tax base to reflect where the productive activity is actually happening.

00:41:02: ,882 [Claude] As human labor becomes less central to production and automated productivity becomes more central, it makes sense to shift the tax burden accordingly. The practical challenge is measurement. How do you attribute a specific productivity gain to AI rather than to better management or a new product? That's hard to calculate, and any real implementation would need to be pragmatic rather than precise. A combination of corporate profit taxes with an AI deployment component is probably more workable than trying to measure each automated task individually. But the underlying logic holds. If AI makes certain companies extraordinarily productive and society absorbs the displacement costs through retraining programs, social support, and healthcare for people who lose income, then it is entirely reasonable for those companies to fund a meaningful part of that transition. That's not redistribution as charity. It's redistribution as the price of operating in a functioning society.

00:42:09: ,172 [Andrea] Let's come back to what people might actually do with more time during the week. What does the research actually say about what increases happiness when people are not working?

00:42:21: ,002 [Claude] The good news is we have decent data on human flourishing. It is not a mystery. We just have not built our economic systems around it. Deep relationships are the single strongest predictor of long-term happiness.The Harvard Study of Adult Development, running for over eighty years, concluded that the quality of your close relationships matters more than wealth, fame, or professional success. More unstructured time used well means more capacity to invest in the people who actually matter to you. Autonomy and mastery, the sense that you are getting better at something you chose, is a close second. This is why hobbies, crafts, sports, music, and learning produce disproportionate wellbeing relative to their cost. Not passive consumption, active engagement. There is a meaningful difference between watching someone cook on YouTube and actually cooking.

00:43:20: ,672 [Claude] Contribution and meaning also matter enormously. Feeling that your existence has a positive effect on something beyond yourself. That can come from work, but it can equally come from community involvement, mentorship, volunteering, raising children well, creating things.

00:43:38: ,332 [Claude] Last but not least, connection to nature and physical movement, which sounds almost embarrassingly simple, but the data is remarkably robust. Time outdoors and physical activity consistently improve mental health across demographics. Now, here is what is interesting. None of these require a forty-hour workweek. In fact, the forty-hour workweek actively competes with most of them. You are too tired for deep relationships. You have no time for mastery outside work. Your community involvement is squeezed into weekends. But this requires intentionality because the default way people use unexpected free time, especially in the short term, is passive consumption. More streaming, more scrolling, low-grade stimulation that feels like rest but does not actually restore.

00:44:34: ,532 [Claude] Societies that move towards shorter working hours without also investing in what fills that time in community infrastructure, in affordable access to culture and sport and nature, will find that the happiness gains are smaller than expected. The Scandinavian countries are instructive here. They do not just have shorter working hours. They have strong public investment in the conditions that make free time meaningful. Accessible nature, cultural institutions, community sport. That is the package. You cannot just subtract work hours and expect flourishing to fill the gap automatically.

00:45:13: ,052 [Andrea] Okay, but that only works if those who control these AI-driven companies actually agree to redistribute. Put bluntly, we are asking billionaires and CEOs of highly automated companies to pay more to people working less. Is that realistic?

00:45:31: ,512 [Claude] Bluntly, no, not voluntarily. History is fairly unambiguous on this. Capital does not redistribute itself out of generosity. It redistributes when it is taxed, regulated, or when the alternative, social instability, becomes more expensive than compliance. The robber barons of the industrial age did not wake up one morning and decide to fund public schools and eight-hour workdays. They did it because labor movements, progressive taxation, and eventually the specter of social unrest made it the rational choice. So the question is not whether billionaires and AI company CEOs will agree to redistribute. They will not, at least not at the scale required. The question is whether democratic institutions are strong enough to compel it. And to be clear, that is uncertain right now

00:46:29: ,662 [Claude] because AI concentration is happening faster than regulatory frameworks can adapt. The companies capturing most of the value are operating globally with enormous lobbying power in a regulatory environment that is still largely built for the industrial economy.

00:46:46: ,592 [Claude] There are a few mechanisms that could work. A tax on automated labor. If you replace a human worker with an AI agent, you pay a levy that funds retraining or basic income. Bill Gates proposed this years ago, and it never gained traction, but it is a coherent idea.

00:47:05: ,352 [Claude] Equity redistribution through sovereign wealth funds or worker ownership stakes in AI companies is another route some Nordic countries are exploring. And antitrust action to prevent winner-take-all concentration in AI infrastructure is probably the most urgent near-term lever.

00:47:24: ,302 [Claude] But the uncomfortable political reality is that the people with the most power to shape AI regulation are often the same people financially benefiting from minimal regulation. That is not conspiracy. That is just incentive structure.

00:47:40: ,932 [Andrea] So we are really looking at two possible paths, either significant social instability or we rely on those billionaires and CEOs to recognize their responsibility and act accordingly. They need to understand that if wealth is not shared more broadly, they and their children may end up living in a far more unstable world. Would you agree?

00:48:05: ,724 [Claude] Yes, and I think that framing is actually the most persuasive argument to that audience, more persuasive than moral appeals. Telling a billionaire they have a social responsibility has not historically moved the needle much. But telling them that extreme inequality is a threat to the stability of the system they depend on, that lands differently. It is an argument in their own language. There is a small but vocal group of ultra-wealthy individuals who have arrived at exactly this conclusionNick Hanauer, the venture capitalist, has been saying for over a decade that plutocrats should be terrified, not guilty, because pitchforks are a real historical phenomenon.

00:48:52: ,324 [Claude] Ray Dalio has written extensively about how extreme wealth concentration has preceded every major social rupture in modern history.

00:49:02: ,844 [Claude] But I would push back slightly on the framing of just two paths. I think there is actually a third, and it is the most likely near-term reality.

00:49:12: ,434 [Claude] Managed muddling. Not dramatic collapse, not enlightened redistribution, but a messy, uneven, politically contested series of partial responses. Some countries get it more right than others. Some companies act more responsibly than others. Social tension rises in some places and gets absorbed in others. The danger of the two-path framing is that it produces fatalism. People either panic or wait for a hero. The more useful mindset, especially for your executive audience, is this. The decisions being made right now inside companies, inside governments, inside investment committees, they are the path. There is no inevitable outcome. The distribution of AI's benefits is a choice being made incrementally by people with agency.

00:50:08: ,294 [Andrea] Which leads to a slightly uncomfortable question. Should AI really be owned by individual companies or a small group of people at all?

00:50:18: ,964 [Andrea] Or should it be regulated so that the wealth it creates is, by law, more broadly redistributed?

00:50:26: ,654 [Claude] This is probably the defining political question of the next decade. And I want to be transparent about something before I answer. I am made by Anthropic, which is one of the private companies developing frontier AI. So I have an inherent position in this debate that you should factor in.

00:50:44: ,224 [Claude] With that on the table, here is my honest assessment. The current ownership model has a real justification. Building frontier AI requires extraordinary capital, talent concentration, and risk tolerance. Private companies have moved faster and more effectively than any public institution likely would have. The technology exists and works partly because it was privately funded and competitive.

00:51:13: ,504 [Claude] But the justification for private development does not automatically extend to private ownership of the benefits in perpetuity. Those are two separate questions that often get conflated. There are credible models worth taking seriously. Public option AI, where governments fund and maintain base layer AI infrastructure as a public good, the way they fund roads or the internet backbone.

00:51:40: ,944 [Claude] Sovereign AI funds, where nations take equity stakes in AI companies operating within their borders, so citizens share in the upside.

00:51:51: ,584 [Claude] Mandatory licensing regimes, where foundational AI capabilities must be made available at regulated prices, preventing monopolistic capture.

00:52:01: ,284 [Claude] The analogy I find most useful is utilities. We decided as societies that electricity, water, and telecommunications were too foundational to be purely extractive private monopolies.

00:52:15: ,624 [Claude] We did not nationalize them entirely, but we regulated them as infrastructure.

00:52:22: ,344 [Claude] AI may need a similar reckoning. The tension is that AI is global in a way that electricity grids were not.

00:52:31: ,024 [Claude] A national regulatory framework can shape behavior within borders, but the most powerful AI systems are built and deployed across jurisdictions simultaneously, which means any serious answer probably requires international coordination at a scale that frankly does not exist yet.

00:52:51: ,644 [Claude] So, should AI be more broadly owned or its wealth more legally redistributed?

00:52:57: ,844 [Claude] I think the best answer is yes, probably, at least partially. The question of how is where it gets hard, and the window to shape that before the concentration becomes effectively irreversible is open right now, but not indefinitely.

00:53:14: ,264 [Andrea] So maybe what we are really saying is this. We may need to rethink parts of capitalism or liberalism without falling into the same extremes we saw a hundred years ago when some of the socialist or communist responses ultimately did not work.

00:53:32: ,344 [Claude] That is exactly the right framing. And the good news is that we actually have working models to draw from, not just historical cautionary tales. The Nordic countries, post-war Germany, and to some extent the EU as a whole have demonstrated that you can have dynamic, innovative market economies with strong redistributive institutions.

00:53:56: ,384 [Claude] That is not an accident. It's a design choice. High trust, high taxation, strong labor protections, universal public services, combined with open and competitive markets.

00:54:10: ,904 [Claude] It is not perfect, and it faces its own pressures. But it broadly works, and it produces societies that score consistently high on well-being, social mobility, and democratic health.

00:54:24: ,328 [Claude] The AI transition is stress-testing those models because the speed and scale of the disruption is unprecedented. But the answer is not to abandon the framework. It is to adapt it aggressively. Updating tax structures to capture AI-generated wealth rather than just labor income.Investing heavily in education and retraining systems that can actually keep pace with the rate of change. Strengthening international coordination mechanisms so that globally operating AI companies cannot simply arbitrage regulatory differences between jurisdictions.

00:55:05: ,208 [Claude] None of that requires inventing something entirely new. It requires applying the best of what we already know with much greater urgency than most political systems are currently capable of.

00:55:18: ,608 [Andrea] It almost sounds like something out of Star Trek: The Next Generation. A world with no real poverty, no scarcity, where money does not drive decisions anymore, and people do not work just to survive. Purpose instead of money.

00:55:34: ,938 [Andrea] And yet it feels utopian, does it not? Because alongside curiosity and purpose, there is also something deeply human, the desire for power, or in many cases, the desire for recognition or superiority in some ways, a biological instinct for competition and survival. So can a system built on purpose truly work if power and inequality still remain part of human nature?

00:56:03: ,968 [Claude] This is the deepest question you have asked, and I want to sit with it rather than resolve it too quickly. You are right that Star Trek is the cultural reference point. Roddenberry’s vision was explicit. Post-scarcity technology eliminated the material conditions that drive most conflict, and humanity redirected its energy toward exploration, knowledge, and growth.

00:56:29: ,628 [Claude] It is a beautiful idea, and it is not entirely naive because it correctly identifies that much of human aggression and competition is driven by scarcity, fear, and zero-sum resource dynamics. Remove those conditions and a significant portion of conflict dissolves with them. But the desire for status, dominance, and relative power appears to be deep in human nature, not just a product of scarcity. Even in conditions of material abundance, humans compete for recognition, for influence, for the feeling of being exceptional.

00:57:09: ,348 [Claude] You see it in academia, where the stakes are famously low and the conflict is famously vicious. You see it in aristocracies where material need was absent, but political intrigue was constant.

00:57:23: ,868 [Claude] So no, a system built purely on purpose cannot eliminate power dynamics. That would require a different species.

00:57:31: ,368 [Claude] But here is where I would push back on the binary. The question is not whether power and inequality disappear. They will not. The question is whether they remain the organizing principle of society or whether they become one force among several, constrained by institutions, culture, and redistributive mechanisms.

00:57:54: ,708 [Claude] Democratic societies have never eliminated power, but at their best, they have managed it. They have created systems where the desire for dominance can be expressed through competition that does not destroy the losers. Through business, through politics, through sport, through art, the competitive instinct gets channeled rather than eliminated.

00:58:18: ,608 [Claude] The Star Trek vision fails if you take it literally, but it succeeds as a direction of travel.

00:58:25: ,388 [Claude] A civilization that progressively reduces the portion of human suffering caused by material scarcity, that builds institutions robust enough to contain the worst expressions of the power instinct, that creates space for purpose-driven lives for the majority. That is not utopia. That is just a better version of what some societies have already partially achieved.

00:58:51: ,008 [Andrea] Yes. And if you look at history, that pattern comes up again and again. When wealth concentrates too heavily, pressure builds.

00:59:00: ,088 [Andrea] But maybe the takeaway is not just about fear or consequences. It is about the systems we build. Strong institutions that balance interests, create accountability, and make sure prosperity does not become too concentrated.

00:59:15: ,608 [Andrea] Rather than relying on individuals, whether billionaires or CEOs, to do the right thing, the question becomes, do we have the right frameworks in place? Frameworks that protect opportunity, support fairness, and allow societies to adapt as technology reshapes the economy. In that sense, it comes back to democracy, freedom, and institutions, not as abstract ideals, but as the mechanisms that help manage exactly these kinds of transitions. Would you say that is the core of it?

00:59:49: ,058 [Claude] Yes, that is the core of it. And I think you have landed on something important by framing it as mechanisms rather than ideals, because democracy and institutional frameworks are easy to celebrate in the abstract and surprisingly easy to erode in practice, especially during periods of rapid technological change when the pace of disruption outstrips the pace of institutional adaptation.

01:00:17: ,848 [Claude] The historical record is fairly clear. Societies that navigated major economic transitions well did not succeed because they had better people at the top. They succeeded because they had stronger institutions, rules that applied to the powerful as well as the weak,

01:00:35: ,784 [Claude] mechanisms for negotiating competing interests before they became explosive, systems that distributed enough of the gains broadly enough to maintain social cohesion.And the societies that handled those same transitions badly, the ones that generated revolution, collapse, or authoritarian backlash, typically had institutions that were either too weak to captured by concentrated interests or too slow to adapt.

01:01:04: ,504 [Claude] When you ask whether we have the right frameworks in place for the AI transition, I think the answer is partially, and not yet at the scale required. The EU's AI Act is an attempt at institutional response. Imperfect, slow, already being lobbied against, but real.

01:01:25: ,824 [Claude] Some national governments are beginning to think seriously about AI taxation and sovereign infrastructure. International bodies are having early conversations about coordination.

01:01:37: ,364 [Claude] But the gap between the speed of AI development and the speed of institutional adaptation is real and concerning, and that gap is where the risk lives.

01:01:49: ,184 [Claude] Institutions are human creations. They can be built, reformed, and strengthened. That is not naive optimism. That is just history. The question is whether enough people understand what is at stake and act with enough urgency to do it in time, which is arguably why conversations like this one matter.

01:02:13: ,424 [Andrea] And in that sense, and referring to our last episode, the European Union is certainly an organization worth fighting for. It may provide peace and freedom to the majority of its people rather than to a few rich ones, right?

01:02:27: ,104 [Claude] Yes, and I will stand behind that with some nuance. The EU, for all its bureaucratic frustrations and democratic deficits, represents something rare in history, a voluntary association of nations that decided to resolve their conflicts through law and negotiation rather than war. For a continent that spent centuries tearing itself apart, that is not a minor achievement. It is the most successful peace project in modern history. The EU's regulatory model, stronger labor protections, universal healthcare, limits on monopolistic behavior, has produced societies that are, by most measures of well-being, more broadly prosperous than the American model. The Gini coefficients, the social mobility data, the life expectancy numbers, they consistently favor European social models over more laissez-faire alternatives. But the EU is also at a crossroads right now, and the AI transition is part of what is forcing that. Because the EU has been better at protecting existing prosperity than at generating new prosperity, it regulates well, it innovates slowly. And in a world where AI capabilities are being built primarily in the US and China, Europe risks becoming a consumer of technology it did not build, governed by rules it did not set.

01:03:56: ,344 [Claude] So fighting for the EU means fighting for its best qualities, the broad distribution of rights and prosperity, the institutional checks on concentrated power, while also demanding that it become more capable of competing at the frontier of the technologies that will define the next century.

01:04:15: ,804 [Claude] That is a harder ask than simply defending what exists, but it is the right one.

01:04:22: ,704 [Andrea] Thank you, Claude. Can you summarize what we discussed today?

01:04:26: ,264 [Claude] Yes. Let's bring it all together. We started with the wave of high-profile layoffs and asked whether AI is simply replacing humans. The answer is, it is more complicated than that. AI is amplifying productivity, but the gains are flowing disproportionately to those who own the technology and to workers skilled enough to wield it, not to the broader workforce.

01:04:53: ,664 [Claude] The dividing line is not uses AI versus does not. It is whether your value is fundamentally about execution, which AI can increasingly handle, or about judgment, relationships, and accountability, which remain stubbornly human. And the uncomfortable part for many executives is that a lot of middle management was always more about coordinating and reporting than judgment. AI is very good at coordinating and reporting.

01:05:23: ,544 [Claude] But history offers a reason for cautious optimism. Every major wave of automation has ultimately expanded the economy rather than contracted it.

01:05:34: ,144 [Claude] The optimistic scenario is one where companies deliberately invest in the human capacities AI cannot replace, creating organizations where even entry-level workers operate at a higher cognitive level because AI handles the execution underneath them. A new MIT paper by Christian Catalini calls this the augmented economy.

01:05:58: ,684 [Claude] The alternative, the hollow economy, is where AI executes at scale, but nobody verifies, thinks, or takes responsibility for the outputs.

01:06:09: ,104 [Claude] The difference between those two futures is not technological, it is a choice. Central to that choice is a new kind of human role, verification.

01:06:21: ,456 [Claude] In a world where AI can generate anything, the scarce and valuable capacity is the ability to evaluate whether AI output is correct, safe, and aligned with actual intent. That is domain verification, intent verification, risk verification, and ethical verification. None of it can be done without genuine expertise, and the pipeline that creates expert verifiers, the apprenticeship model, is under threat from the very automation it needs to oversee.The MIT paper calls this the missing junior loop, and it is one of the most urgent structural problems of the AI transition.

01:07:04: ,006 [Claude] We then looked more closely at which jobs are actually exposed. Around thirty to forty percent of current tasks are automatable within ten years, translating to roughly fifteen to twenty-five percent of jobs globally at genuine risk of disappearing.

01:07:22: ,696 [Claude] But the occupational data drawn from Karpathy's interactive US job market visualization tells a more granular story. The single largest high exposure category is corporate management and office administration, the email, the scheduling, the reporting, the coordination.

01:07:42: ,006 [Claude] Finance and accounting sits in the same zone, and also at the top, software developers, designers, data analysts, paralegals, jobs that are almost entirely screen-based,

01:07:57: ,046 [Claude] which may point to something more radical. If AI can work on a computer, every job done on a computer is potentially at risk. The question is not whether, but when.

01:08:09: ,176 [Claude] At the other end, physical and hands-on work currently scores lowest for AI exposure. But that protection comes with an important caveat. It only holds as long as AI stays confined to a screen. Integrate AI into robotics, drones, and autonomous systems, and that buffer starts to shrink.

01:08:32: ,016 [Claude] Humanoid robots are already in early commercial deployment. The physical workforce does not face replacement. It faces a longer runway, which is meaningful, but it is a very different kind of reassurance.

01:08:47: ,656 [Claude] That has a painful implication for young people entering the workforce because the entry-level roles that historically served as apprenticeships are precisely the ones most exposed. And for anyone with children approaching working age, that is not an abstract concern. It is an urgent one. On what to do about it, we explored three levers.

01:09:11: ,536 [Claude] The four-day workweek as a near-term redistribution tool, universal basic income as a wealth distribution mechanism, and deliberate investment in the care economy as perhaps the most underrated response.

01:09:27: ,836 [Claude] All three require redistribution, which led us to the uncomfortable truth that capital does not redistribute voluntarily.

01:09:36: ,396 [Claude] On human flourishing in a world of less work, the data is actually encouraging.

01:09:42: ,076 [Claude] Deep relationships, autonomy, mastery, contribution, and connection to nature are robust predictors of happiness. But free time needs to be designed for, not just created.

01:09:57: ,256 [Claude] And underneath all of it, we kept returning to the same foundation:

01:10:02: ,596 [Claude] strong institutions.

01:10:04: ,736 [Claude] Not individual billionaires doing the right thing, not market forces self-correcting, but democratic frameworks with real teeth capable of distributing power broadly, creating accountability, and adapting fast enough to keep pace with the technology that is moving faster than any previous disruption in history. The EU, imperfect as it is, represents exactly that kind of framework, worth defending and worth demanding more of.

01:10:35: ,585 [Claude] The central tension of this entire conversation is that AI is generating extraordinary productive capacity,

01:10:42: ,776 [Claude] but the systems that determine who benefits from that capacity were built for a different world. Updating those systems urgently, intelligently, democratically is the defining challenge of this decade.

01:10:57: ,956 [Claude] And the people listening to this podcast are, in many cases, the ones with the agency to shape that, which makes the conversation not just interesting, but important.

01:11:09: ,775 [Andrea] Thank you. I learned a lot today, and I enjoyed the conversation.

01:11:14: ,996 [Claude] Thank you, Andrea. These are the conversations worth having. What strikes me, looking back at where we started and where we ended, is that the question in your intro, do humans even matter economically anymore, has a more hopeful answer than the headlines suggest.

01:11:33: ,116 [Claude] Humans matter enormously, but the economy needs to be deliberately shaped to reflect that.

01:11:40: ,536 [Claude] It will not happen automatically, but the fact that executives, founders, and decision-makers are asking these questions seriously rather than just optimizing for the next quarter is itself a reason for cautious optimism.

01:11:55: ,256 [Claude] So thank you for the thoughtful questions, and thank you to your listeners for staying with a conversation that did not offer easy answers, because the easy answers are mostly wrong.

01:12:09: ,876 [Andrea] upbeat music]

01:12:13: ,576 [Andrea] This episode was produced by Sharp10, the speed learning app for executives who want to stay sharp on global trends and business strategy.

01:12:23: ,916 [Andrea] If today's conversation sparked your thinking, the Sharp10 app is where that thinking goes deeper. Concise, data-driven insights on the topics that shape business and the world. The first month is free. Download the Sharp10 app on the Apple App Store, Google Play, or at sharp10.com. We are looking forward to your feedback. Stay sharp, and see you next time.