Anthropic’s Report on the US Economy in 2030 as AI Advances

Asking About the Economy of 2030 Means Asking About Reality Today

Today is September 14, 2026, and I want to read the Anthropic report I mentioned last Friday with you. It examines how advances in AI could change the US economy by 2030, roughly four years from now. This is not a distant future decades away, but a period we will soon encounter. Nor does the report offer a single prophecy declaring that things must turn out a particular way. It presents scenarios: what economies might emerge if AI advances modestly, substantially, or to an extreme degree.

Drawing on a technical report and an economic model, Anthropic has published a web report that ordinary readers can follow. It does not cover every problem in the American economy and is quite compressed. In some respects, it is excellent. I have criticized many reports about AI and employment, but this one contains things worth praising. At the same time, there is no shortage of problems to criticize. I want to discuss both.

Earlier reports told us to have children study mathematics or engineering because AI was advancing. Others said AI would not have much impact, and we were endlessly told that new jobs would appear. AI company CEOs and reports from prominent companies kept repeating that nonsense. This report, however, at least addresses the problem I have continued to raise: as AI advances, it can accelerate the process by which the poor get poorer and the rich get richer.

That is why I chose “AI Advances Accelerate the Rich Getting Richer and the Poor Getting Poorer” as the main headline for today’s thumbnail. To me, this is the report’s most important point. You may read it differently, of course. That is precisely why we should look at what it actually says together. A forecast for America in 2030 is not merely an American issue. What will happen to jobs in Korea? Who will receive the world’s wealth? How will we respond? Those questions follow as well.

Why Is an AI Company Speaking Before Governments and Intellectuals?

Before entering the report, there is something I want to ask. Shouldn’t economic ministries, governments, economists, university professors, intellectuals, and influential YouTubers in Korea and America be discussing this? Why aren’t they speaking? Change is unfolding before our eyes. Why isn’t there a serious debate?

In my view, OpenAI and Anthropic are competing for the lead in frontier artificial intelligence. Anthropic has now issued this report. Does that mean only Anthropic could have produced it? Do people in Korea have no eyes? Can American economists and the government, or the Korean government, legislators, and professors, not see? This is not something that requires access to an extraordinary secret.

I give Anthropic some credit. Earlier reports contained so much nonsense that simply acknowledging the possibility of a widening wealth gap gives this report value. But that is where my praise stops. We must recognize reality to make predictions, and we need predictions to discuss government policies and social and economic responses. How can we respond while closing our eyes to the present and failing to see the future?

The Economic Model’s Conditions and Its Stated Purpose

Let us read a Korean translation of the economic scenarios on Anthropic’s website. When it says “our economy,” remember that “our” refers to the United States. The report begins with questions: will AI bring unprecedented growth, widespread unemployment, or some other outcome? It explains that a model was created to explore effects on American jobs, growth, and unemployment.

The separate technical report contains a great deal of mathematics. It is an economic model, so mathematical expressions and complicated equations are to be expected. It establishes a framework for calculating what outcomes might follow from particular conditions. I am not proposing to examine all that mathematics today. I want to follow the logic of the public-facing report and consider what it assumes and what it leaves out.

The screen presents three scenarios in blue, green, and orange: limited AI impact, substantial impact, and rapid advancement. One statement deserves attention immediately: the changes should benefit society as a whole, including workers. What if AI advances but only certain people gain? That question is present from the outset.

The report distinguishes an economic index measuring current AI use from a scenario explorer looking ahead. Its argument is that a clearer view of the future can help us take measures that allow everyone to benefit. The proposition that society as a whole can become wealthier is obvious enough. Agricultural and industrial revolutions, electricity, the internet, science, and technology have all increased total production. The question is whether that wealth reaches everyone. That basic position is quite close to mine, and it deserves praise.

More Things to Do Are Not the Same as More Jobs

The report sees the economy as a collection of tasks people perform. AI can help people do some tasks better or faster, automate others, and leave still others unaffected. It may also create new tasks. I do not deny that new activities can emerge. There are already solo entrepreneurs, and although I am still the same YouTuber, the range of things I can do alone has expanded enormously compared with one, two, or three years ago.

AI has helped bring that change about in a short time. I can now do things I could not previously manage by myself. But does that automatically mean wealth is being redistributed? Being able to do more myself is not the same as there being more places for people in the labor market. We need to keep that distinction in mind throughout this report.

It is easy enough to understand how AI assistance raises productivity. But when someone adds, “therefore new jobs will be created,” we must ask what is being created and how many people it requires. If many existing tasks disappear and only a little checking remains, calling that a new job does not adequately describe reality.

What If AI Changes a Nurse’s Working Day?

The report uses a nurse’s day as an example: making rounds to check on sick patients, drawing blood, triaging arrivals, recording vital signs, and preparing ward supplies. A single occupation contains many tasks. It explains that these tasks were classified using the US Department of Labor’s occupational framework.

The work changes over time. Charts were once written by hand; now they are entered on computers. The report gives bathing a patient as an example of something AI cannot do, suggesting that certain tasks still require humans. But won’t physical AI change that too? Something remaining a human task today does not mean it will remain exclusively human in the future. I will return to this point.

Other tasks can be augmented by AI: remotely monitoring patients, planning nursing schedules across shifts, and preparing discharge instructions. Much of writing, planning, and managing can involve AI assistance. Recording vital signs or ordering supplies may be fully automated. A large portion of the nurse’s existing duties then disappears.

The report says new tasks will emerge, such as checking whether AI triaged a patient correctly or reviewing a proposed treatment plan. I can agree with some of that, but frankly I consider parts of it imaginary. Does this create work requiring many additional people, or does it leave a tiny remainder after drastically reducing existing duties? Calling something checking or reviewing does not restore the labor that disappeared.

Nor does this happen to just one nurse. It changes the entire set of tasks countless nurses repeat every day. The same person can accomplish more in the same time, increasing productivity. If the economy’s size represents all the work performed by people, machines, and software, then the way AI changes these tasks will also change GDP and workers’ incomes.

From Thirty Trillion Dollars to Three Paths for 2030

The report starts with a US economy of roughly $30 trillion in 2025 and charts its path to 2030. The lowest case reaches approximately $34 trillion, the substantial green scenario about $36.3 trillion, and the highest about $44.4 trillion. Starting from the same point, the destination differs according to how AI develops and is adopted.

Technical capability is not the only important variable. Does AI assist with work or automate it completely? How much does it improve productivity? How quickly do workers and firms adopt it? How many new tasks appear? These conditions jointly change economic output and the labor market. That is why the report separates the three cases.

One condition associated with the extreme case is recursive self-improvement. This concept appears when discussing rapid AI advancement that sharply increases production. Why introduce a technical idea about self-improvement into a discussion of the economy just four years away? Because it can change the speed of research and development itself.

Recursive Self-Improvement and the Idea of the Singularity

You have probably heard the word singularity. It imagines a point beyond which AI becomes so overwhelmingly intelligent that humans can neither keep pace with its development nor readily comprehend its abilities. Various explanations have been proposed for how such a point might be reached. RSI is one of them.

RSI means recursive self-improvement. I checked the English during the broadcast and found that I had written “recursive” correctly. I even laughed: “How did I get that right for once?” You can understand recursion as calling oneself again. Self-improvement means improving oneself. What happens if AI makes itself better and that improved AI can then improve itself again?

Humans currently develop AI, but they already receive considerable AI assistance during development. Setting aside whether this qualifies as full recursive self-improvement, I believe OpenAI, Anthropic, and Google have already entered parts of that process. They apply their latest models to developing the next models. AI is used in mathematics, algorithms, methodologies, kernels, semiconductor development, and coding. Using AI to build the next AI is already happening.

I also connect this with Anthropic’s recent suggestion that companies monitor one another and establish tools and organizations to slow development somewhat. Capabilities are advancing so quickly that safety procedures and assessment of social effects struggle to keep up. Developers themselves are sensing that introducing elements of self-improvement could sharply accelerate progress.

There are stages, of course. Fully recursive self-improvement would mean AI continually advancing itself without a human lifting a finger. That is frightening. Semiconductors and physical energy also impose conditions, but even in software alone, I think neural network architecture and inference methods remain at an extremely early stage. Using AI to improve their efficiency leaves room for rapid progress. The extreme scenario assumes this kind of self-improvement and fast adoption.

Why I Think the Modest Scenario Already Underestimates Reality

Now consider the lowest scenario. It depicts only a tiny difference between little AI and no AI. This is where I already find it difficult to agree. AI existed at the starting point, and today’s systems already demonstrate extraordinary abilities. Yet the picture suggests merely gradual effects within the ordinary historical range of new technologies. To me, that looks like a lie that turns away from reality.

In the substantial green scenario, AI can perform half of knowledge work by 2030. Yet GDP rises only from around $30 trillion to $36 trillion. AI can perform half of knowledge work, and that is the scale of the change? I consider that excessively conservative. It is why I find the chart absurd.

Reading further, we encounter the qualification that capability does not mean universal adoption and that most knowledge work continues without AI. Even so, I cannot accept the assumption that things will remain at that level four years from now. I think nearly 70 to 80 percent of human knowledge work will then be replaceable. That is why I use the strong phrase “complete bullshit” about this modest picture. Given present changes and the pace of development, is it reasonable to draw such a low trajectory?

The substantial scenario says the economy grows twice as fast as usual while knowledge workers’ wages do not rise. How could their wages rise? There may be additional token costs for AI, but if one person can do twice as much work or more, human wages are more likely to fall. It says other workers’ wages rise, however, and I disagree with that too. I will examine why separately.

What If AI Also Eliminates the New Jobs Supposedly Supervising It?

In the extreme scenario, AI is more productive than humans at most knowledge work. It writes better reports and books than you, produces better pictures, handles mathematical problems and proofs better, and conducts better research. It does not merely assist with half of human tasks; it autonomously performs most knowledge work.

The explanation then says that it creates almost no new knowledge work for humans. Today, after AI does something, a person judges, monitors, checks, and directs it. But what if AI also performs that judgment and checking better than a human? Employment can no longer be guaranteed by insisting that someone will always be needed to check its work.

This scenario describes annual GDP growth reaching 15 percent as AI spreads, with the economy doubling every four and a half years. Society becomes much wealthier, but the number of people engaged in knowledge work drops sharply, and unemployment exceeds levels associated with an ordinary recession. This is not a temporary shortage of work caused by an economic downturn. People lose jobs because AI replaces the labor itself. I read that as a collapse of the labor market.

Survey respondents expecting the substantial scenario does not make that path correct. People’s sense that things will probably turn out a certain way is not evidence of the actual future. More important to me than the GDP figure is the position workers occupy within that expanded production.

The Economy Grows, but Who Gets the Money?

The report itself says growth is not the only concern. We must consider how much workers receive in pay and how many must find new jobs. Even if national GDP grows and society becomes wealthier overall, much of the prosperity may accrue to the resources and technologies that produce wealth rather than to workers.

Who owns those resources and technologies? The rich do. If AI can use resources and energy to make cars, develop software, conduct research, and run industry, production can increase enormously. Yet fewer people are needed. The gains flow to those who possess the money, resources, and technology that made production possible. They go to the rich.

I see no need to wrap this in complicated language. If society becomes wealthier but prosperity goes mainly to capital owners, the rich get richer and the poor get poorer. This report says AI advancement can accelerate that widening gap. That is why rising GDP alone is no reason to applaud.

Where Do Displaced Office Workers Go?

The report says job reallocation and unemployment remain within historical ranges in most scenarios, while unemployment can surge to historic levels in the extreme case. It also explains that many job seekers normally find another job relatively quickly. But doesn’t that presuppose that there are jobs to move into?

It is correct that changing occupations completely is difficult and finding a new position can take a long time. Suppose someone who used Excel and produced documents in an office is dismissed. The company now finds AI better at reports, market research, Hangul word-processing documents, summaries, meetings, forecasts, and Photoshop. Most of what that person did at a computer has been replaced.

Can that person immediately move into physical work? Is it easy for someone who handled software and documents to switch abruptly to work requiring different skills and experience? Of course not. This cannot be disposed of with the phrase “occupational transition.” People still have to live while that transition takes place.

Is the Explanation for Higher Wages in Other Occupations Sufficient?

The report says average wages rise in all three scenarios, with gains concentrated outside knowledge work. People working in offices and laboratories, coding, editing, drawing, or using Photoshop and Excel see stagnant or falling wages. Meanwhile, physically oriented occupations such as plumbing, construction, and cleaning may receive higher pay.

Construction is the example. If AI speeds up infrastructure design and permitting, more construction projects follow, increasing demand for construction workers and raising their wages. Knowledge workers’ wages barely change in the substantial scenario and fall by more than 10 percent by 2030 in the extreme one. I understand the argument that replacing humans with AI pushes knowledge workers’ wages down.

But are wage gains in other occupations really that simple? If prices rise too, does an apparent pay increase mean life has improved? We need to distinguish nominal increases from actual living conditions. Has the report adequately considered robots entering non-knowledge work and unemployed people flooding into those occupations? Without those two questions, wage projections can become excessively optimistic.

Labor’s Share, Capital’s Share, and the Distribution of a Larger Pie

A company needs money and people. It raises capital by selling shares, borrowing, or obtaining loans. People work to provide services, manufacture shoes or semiconductors, and sell the results. Who receives the proceeds? Some go to those who supplied funds and own shares; some go to the workers. That is the distribution between labor and capital.

The report explains that roughly 60 percent of each dollar produced by today’s American economy goes to workers and 40 percent to capital: approximately a six-to-four relationship. But when AI automates more tasks, fewer people are needed. A larger proportion of the same dollar may then go to capital.

The pie itself can grow. Yet if capital’s share grows faster within it, we cannot simply say workers are becoming wealthier too. Through what route do workers receive income when the labor market collapses? The gains return to those who invested money and own resources. The report also explains that capital’s share can grow more even when all workers’ wages increase to some degree.

As capital becomes useful in more places, demand and prices rise, and more growth accrues to owners. Describing AI advancement only as a productivity gain conceals this ownership problem. I keep discussing the rich getting richer and the poor getting poorer because who has the rights and means to obtain goods matters just as much as how many goods are produced.

Can We Forecast Manual Work While Leaving Out Physical AI?

The limitations section says the model does not include a scenario involving the development of robots with superhuman capabilities. Physical AI is missing. I say this is one reason to call the report crap. A change is approaching right before our eyes. Is it acceptable to exclude it and then speak as though other occupations are safe?

My example is Hyundai Motor. Consider workers asking that the introduction of robots to automobile production lines be postponed or slowed. Physical AI is not an abstract, distant future. If we exclude robots and then predict falling pay for knowledge workers but rising pay for everyone else, the distinction itself may fail to match reality.

Parcel sorting was also human work. Robots are now beginning to identify packages and determine where they should go. I also saw an example of mackerel processing, somewhere around Norway, I think. People used to separate bones and flesh, cut, wash, and package the fish. Now AI uses cameras to locate the bones and handle the process. From the fish entering to its emerging vacuum-packed, almost everything proceeds automatically with virtually no manual handling.

Why are fewer people needed there? Because artificial intelligence has been combined with physical equipment. I cannot agree with predictions of higher manual wages that completely ignore this. The report promises continuing updates as new evidence appears, but the problem remains: its forecast leaves out things already happening.

Reviewers Also Raised the Problem of Underestimation

The report concludes by introducing criticisms from outside reviewers. One is that the model does not track individual workers and therefore captures the costs of job loss only very roughly. Some think occupations exposed to AI could grow rather than shrink; others see the extreme case more as a thought experiment than a scenario.

Conversely, some say even the modest scenario underestimates changes already visible in the data. That is exactly my point. Companies are already reducing staff because of AI, and the IT industry needs fewer people writing code by hand. Why present a path that looks almost unchanged? Staff reductions happened last year and early this year, and I expect more.

Reviewers also criticize the model for underestimating how much AI will accelerate technological progress itself. The authors acknowledge these as limitations of the current model. That is why I call the report worthwhile yet problematic. Read it yourself if you are interested. If English is a barrier, use Chrome translation or give an AI the address and ask it to translate. You need not struggle to translate it manually. Other tools offer better translations too.

Two Reasons for Praise, Preceded by a Problem of Self-Defense

Let me now organize my assessment. First, the report addresses both positive and negative effects of AI advancement. Productivity rises, but knowledge work loses value, the labor market collapses, and gains may concentrate among those who own money and resources. Presenting both sides in the same report is meaningful.

Second, an organization developing AI has projected the social shocks associated with the pace of advancement. Much of the discussion until now has trumpeted only the benefits. People like Elon Musk say everyone will enjoy abundance. But does higher productivity make everyone rich? What happens when people have no right to receive that wealth? We cannot speak only of abundance without asking these questions.

It cannot be easy for an AI company to publish a report suggesting that better AI could leave many people poorer. In that respect, I think Anthropic is better than the Korean government, legislators, professors, research institutions, and President Lee Jae-myung. At least it has discussed the problem publicly. What are the people called intellectuals doing when they say nothing about what is happening before them?

Yet there is a problem that belongs before these points, as number zero: the publisher’s self-defense. A capitalist AI company cannot abandon its own corporate position. This is why I criticize the rest as looking half like lies. It acknowledges concentrated gains, but I see the report as tilted toward minimizing real shocks and reducing people’s fear.

Blocked Hiring and a Ladder Being Kicked Away

I do not think the report adequately reflects actual unemployment and hiring difficulties. Many people cannot find work even after completing a four-year university degree. Whether in computing, law, or office work, why hire newcomers when existing employees suddenly become two or three times as efficient? Unless the volume of business rises proportionately, more staff are unnecessary.

Have you asked AI to write a report, create images, summarize material, or handle Excel work? It is startling to see tasks that took hours completed almost instantly. One person’s capacity doubles or triples. Inside a company, productivity improvement can mean a reduced need for new hires.

This concerns more than dismissing existing staff. Positions where newcomers can begin a career disappear. Employment doors narrow for new lawyers, doctors, office workers, and programmers. When existing programmers become much more productive, there is less reason to hire new ones. Who will recruit while already busy cutting staff? I call this a ladder-kicking effect.

In forecasting 2026 through 2030, has the report adequately included changes already occurring in 2026? What kind of prediction ignores the present? Even now, while people debate whether we have superintelligence or AGI, people are losing jobs and struggling to enter employment. It should not be portrayed as though serious shocks arise only with extreme advancement.

Autonomous Driving Is Already a Labor Issue

Autonomous vehicles are among the first things to consider when discussing physical AI. Korea has some pilot services and truck trials, but it is harder to see driverless taxis operating routinely nearby as they do in America or China. That may make the change feel less immediate. Elsewhere, however, robotaxis already operate. If replacing taxi drivers is not physical AI, what is?

I now regard autonomous driving as a question of when rather than whether. Apart from people driving as a hobby, I expect AI to replace most driving, including taxis and trucks. We therefore need to ask how many people in Korea earn their living by driving.

When I researched this while writing a book several years ago, the figure was roughly over a million. I mean ground transportation such as buses, taxis, and trucks, excluding maritime work. If employment on that scale is affected, can a picture of lower knowledge-worker wages and everyone else doing fine explain it? My criticism that the forecast barely considers physical AI applies here too.

Development Accelerates Even Before Full RSI

I also object to treating recursive self-improvement only as a condition of an extreme future. Companies, including Anthropic, use AI across numerous parts of AI development. Looking into it, I found the use much more active than I had expected. I saw Google as virtually staking its future on RSI to recover ground in the frontier competition. Anthropic uses current models in subsequent development, as does OpenAI.

That is the context in which RSI appears in discussions of internal audits or checks on whether the current stage is dangerous. Effects arise even before perfect self-improvement, where AI develops everything independently of human involvement. Applying AI to one mathematical problem, one software bottleneck, or one part of a development method speeds up what follows.

Whether it strictly qualifies as recursive is a separate question. AI accelerating AI development already matters. That is why I strongly oppose presenting the minimally changing blue line as a realistic possibility. A low case that omits the acceleration already under way can begin from the wrong starting point.

How Do Displaced Workers Affect Wages in the Jobs That Remain?

Knowledge workers’ wages fall because their AI competitor is cheaper and better. But where do the people pushed out go? They learn plumbing, enter construction, care for older people, and seek work AI cannot immediately replace. They have to live. They cannot simply remain idle.

The labor supply in the remaining occupations then increases. The report discusses greater demand for those occupations, but has it sufficiently considered the people newly entering them? Thinking in terms of supply and demand, we cannot simply assume their wages keep rising. On top of the difficulty of changing occupations, wages after the transition may not rise as expected.

We must also consider wages’ relative position. What if total production and goods increase, and capital and asset values rise, while my wages stay unchanged? Imagine property prices and living costs climbing without a pay rise. An unchanged number does not mean an unchanged position. I view stagnant knowledge-work wages as a relative decline too. Even if other wages appear to rise, we must ask how much life actually improves.

Even if wages rise temporarily in some occupations, why assume that lasts? The entire world is racing to develop robots. Will robots not enter construction? Care centers and elder care? Cleaning, parcel sorting, and factories? I expect rapid entry into these fields too. Discussing higher non-knowledge wages without this change is profoundly inadequate.

New Activities and Positions in Wage Employment Are Different

I ask whether even the people promising new jobs after AI still believe what they say. For years, the AI industry and people such as Sam Altman have talked about new jobs. What exactly will appear, and in what quantities? Repeating the claim without grounds for replacing disappearing employment makes it an outright lie.

I am not saying humans will do nothing. I think the concepts of work and hobbies may change. Today we call it work when we are hired by someone, do what we are told, and receive money. I go so far as to describe such subordinate labor as a slave’s work. But if someone comes from a wealthy family owning eight properties, why must they work under another person, watch that person’s reactions, and follow orders? They could live doing what they want.

Under those conditions, a hobby can become work and work can become a hobby. AI can handle necessary tasks people dislike, while people create what they want. But that change is different from companies creating new salaried positions. New activities must not be used to conceal declining employment in the existing labor market.

A Website I Built Alone and a Game I Can Now Make

Enormous productivity gains for individuals and small teams are already happening. I developed a website in a few days that would once have been difficult to build alone. I am the same person, but different tools changed what I could accomplish.

The same applies to games. There was a game I wanted to make in the past but could not. I was stopped by an inability to draw or implement certain parts. Now I can discuss the idea and overall structure with AI and ask for the necessary artwork and functions. I handle what I do well and receive help where I am lacking. I have been thinking about making a game lately, but have not started because I wonder whether that time should go to Buddhist broadcasting instead.

Something that once required more than ten people working for two or three years can potentially be accomplished in a short time by one person with AI. This kind of activity will increase. But is it a new job in the existing sense? It is not a salaried position under someone else, following instructions and worrying about their approval. It is closer to a solo or small-team venture.

I do not deny the possibility. I benefit from it myself. But expanding individuals’ ability to start businesses must not be equated with an increase in jobs employing large numbers of workers. New activities and ventures will certainly emerge in the short term. Saying that alone resolves the labor-market crisis, however, is a lie.

A Forecast for 2030 That Does Not Look at 2026

To predict the future, first examine what AI changed in 2026 compared with 2025 and 2024. Which tasks disappeared? Which industries are breaking down? Evaluate reality before discussing 2027 and 2028. Drawing a picture that reduces fear while ignoring current changes is not a proper forecast.

Short-form dramas provide another example. Korea is experimenting too, but I see particularly strong activity in China producing large quantities of short dramas with AI and no actors. The efficiency difference from filming with people is substantial. They may still look AI-generated and inferior to human productions. But for how long? Might we soon be unable to distinguish real actors from AI?

Recently, I too have become unable to distinguish well-made AI images from photographs. You may think you can tell from strange hands or obvious errors. Failed images like that exist. But well-made images are now hard to distinguish. Across many fields, replacement of human work is already under way.

That is also why I discuss the shock delivered this year by Astra and Fable 5.1. Having seen those changes, have you imagined six months or a year ahead? And yet we draw such a modest picture four years out? Meanwhile, developers are talking about slowing progress. We can discuss slowing it. But at the very least, governments, intellectuals, and professors who act so clever and important must open their eyes and talk about how this world will change.

Consider Individual Possibilities Alongside Society’s Responsibilities

I believe parts of the labor market are already collapsing. What should individuals do? They can use AI’s productivity to attempt things previously impossible: develop ideas, map the overall design, and create something through dialogue with AI. That can also help them try to earn money from the results. Young people can experiment with many possibilities.

In one respect, this is a good world. What an individual can accomplish has expanded astonishingly. In another, many people struggle amid extreme instability and rapid change. Both are true. I think the old expectation of a stable job in a world that will not change much is gone.

We therefore cannot stop at telling individuals to use AI well. If advancement makes the rich richer and the poor poorer, how do we prevent that? Will we allow only the rich and resource owners to benefit? What must change so that people throughout the world, not only in Korea, can share the gains?

Why Is That Your Resource? We Must Reopen the Question of Sharing

This brings me to ownership. Why is that your resource? Why is that your energy? Even if we cannot immediately abandon all the value created within the illusion of money, we should not take present ownership and distribution arrangements for granted. I think we need new concepts, debate, and discussion, and must move much of society toward sharing.

If Korea’s Constitution needs changing, we should consider a direction founded on sharing resources and land, society, literature and the humanities, technology, and science. Could we recognize individual effort on top of that shared foundation?

Will we allow the gap between rich and poor to widen, or seek a way to live well together? That is the question I leave after reading this report. Society’s ability to decide how the benefits of production are shared matters as much as technology’s ability to increase production.

Why I Leave Both Praise and Criticism

Lately I have tried to finish broadcasts in around an hour, but I have spoken at length again today. I would like to converse when chat messages arrive, yet when it is quiet I keep explaining things alone, so the broadcast style seems to have changed a little. Dao Sunim said it was enjoyable. Please join the discussion if you have questions.

I see this Anthropic report as an important early direct admission that the benefits of AI development can concentrate in capital and the wealthy, accelerating the rich getting richer and the poor getting poorer. I want to praise an AI company for saying so. At the same time, I strongly criticize its naivety and its attempt to forecast without adequately examining reality.

We do not have to choose just one of those responses. Acknowledge a meaningful admission, and examine omitted realities and mistaken assumptions. That is why this deserves our shared attention. May you attain Buddhahood, and may you begin the new week with energy and good days ahead.