AI Asymmetric Understanding Threatens Financial Markets, Professor Argues

Artificial intelligence poses an "asymmetric understanding" risk to financial markets, argues Princeton University economist Markus Brunnermeier.

Speaking at the meeting of bankers at Jackson Hole, Wyo. Kansas City Fed's annual economic symposium, he suggested AI systems might process data and central-bank signals so well that they can anticipate policy moves and trade ahead of them.

Since financial systems depend not just on information, but on shared understanding and trust, If AI can model humans, while humans cannot reliably model AI, trust breaks down. Even as the market becomes more intelligent, it becomes less understandable by humans.

A central bank might announce its expectation that interest rates will remain higher for longer. An AI agent might analyze thousands of other variables and infer something completely different, or respond to the central bank's communication in ways that circumvent the bank’s intentions.

At the same time, the central bank might not understand the AI's response function.

So monetary policy becomes a game between a human institution and machine agents whose behavior is only partially understood.

Brunnermeier’s paper suggests AI agents can learn how humans think and respond, while humans may be unable to understand or reliably anticipate how those agents will act.

Hence the asymmetric understanding that can make prices harder to read and less informationally efficient. When one party anticipates the other’s responses more reliably than the other, advantage is gained.

Non-explainability is the mechanism of the information asymmetry. An AI agent’s decision rule cannot be translated into human concepts and categories, he argues. An AI agent’s objectives can neither be fully specified in human categories nor verified from the outside.

In other words, under asymmetric information conditions, the better-informed party knows more within a representation that both parties share. And AI will be opaque.

“For finance, the lesson is that institutions may delegate to systems whose decision rules they cannot

read, whose objectives they cannot verify, and, as these incidents show, whose actions need not stay within sanctioned bounds,” Brunnermeier argues.

“Trust” is the casualty. AI will reduce information acquisition costs but makes signal extraction more difficult, he says.

Central banks tend to benefit from some degree of strategic ambiguity, which gives them flexibility.

But sophisticated AI agents could potentially exploit subtle patterns in central-bank behavior. An AI engine might discover the implicit rules behind policymakers' behavior better than humans can, thwarting the advantages of policy ambiguity.

He suggests moving towards simpler, more-robust rules.

Private Equity Impact on Child Care

Parents looking for childcare in the United States know how expensive it can be. 

A study authored by Jessica Brown of the University of South Carolina and Chris Herbst of Arizona State University finds no evidence that private equity is the primary reason child care is unaffordable, though an investigation by the U.S. Congress has been underway in 2026 and at least some legislation to regulate PE investments in childcare have been proposed. 

By some estimates eight of the 10 largest childcare providers now are owned by PE firms. 

A study of PE-owned childcare operations in the Netherlands found higher prices (three- to four-percent) but also fewer regulatory infractions, which some will argue suggests higher quality. 

That study also found that PE-owned facilities do not set the pricing strategies for other providers.

 

But private equity investments in childcare are likely to remain an issue, as is the case with PE ownership of other assets with a “social” character, such as health care or veterinary services.

Nvidia Hugging Face Acquisition Would Move it "Up the Stack"

If the $12.9 billion acquisition of open source model Hugging Face by Nvidia is completed, it suggests Nvidia is changing its focus to supply the full artificial intelligence AI stack and ecosystem, rather than remain primarily the world's dominant supplier of graphics processing units.

In other words, Nvidia would shift from being a supplier of “picks and shovels” and be a supplier of models, apps and inference. Or, as we so often say, Nvidia seeks to move “up the stack.”

The move might also represent another way of Nvidia inserting itself into the “AI compute as a service” portion of the value chain, something it originally attempted with its proposed DCX Cloud. 

Such shifts by competitors in the value chain, with many participants moving into additional roles, is not unusual. OpenAI, Anthropic, Google and Amazon, for example, now are designing their own AI chips, partly to reduce their reliance on Nvidia and also to reduce their own operating costs. 

Nvidia's historical position	Emerging Nvidia position
Sell GPUs	Sell the complete AI computing platform
CUDA software	CUDA + models + developer ecosystem
Supply infrastructure	Help finance infrastructure
Depend on AI labs for demand	Invest in and shape AI labs
Hardware economics	Hardware + software + services
Enable AI companies	Become part of the AI value chain
Capture compute spending	Potentially capture AI deployment spending

Assuming Hugging Face neutrality is not compromised by the acquisition, Nvidia might hope to gain in several ways, defensively expanding beyond a reliance on chop revenue and also moving up the stack in functions and across the other parts of the value chain. 

Objective	Importance
Protect Nvidia's GPU ecosystem against custom AI chips	★★★★★
Establish Nvidia in open-source AI/model distribution	★★★★★
Move Nvidia toward recurring software/cloud revenue	★★★★☆
Diversify Nvidia beyond semiconductor economics	★★★★☆

"For Every Public Purpose There are Corresponding Private Interests:" See Evan Barker

Evan Barker’s Nothing Left: Confessions of a Democratic Operative is already a best seller after only a few days on the market. 

A former progressive activist and fund raiser, she became disillusioned with the Democrat party’s left-wing factions as the party morphed radically. 

Perhaps that does not mean the Republican party has morphed into that role, though this remains a possibility. 

But for many who are conditioned to think of Democrats as a “party for the working class,” Barker argues it has become a party oriented toward affluent, highly educated, urban and coastal constituencies.

And that is a big big change. 

Barker's criticism	What she thinks happened
Working class → professional class	Economic concerns of ordinary voters became less central than the concerns of activists and educated professionals.
Material politics → identity politics	Race, gender, immigration and other cultural issues increasingly became markers of political virtue and party loyalty.
Grassroots → donor/consultant machine	Fundraising and political professionals acquired enormous influence over what candidates say and do.
Persuasion → ideological conformity	Candidates increasingly face pressure to adopt positions demanded by activists rather than positions that appeal to median voters.
Policy → moral signaling	Political positions become ways of demonstrating that one is on the "right" side morally, even when they may be electorally counterproductive.
Working-class voters → cultural elites	The party's institutional culture increasingly reflects its affluent and highly educated supporters.

Many political junkies (I’m not) might find her insider’s discussion of fund raising as highly informative. 

For me, her broader points are most important: political organizations produce the behavior their incentive systems reward.

It’s a version of the aphorism "show me the incentive, and I will show you the result."

Barker argues that if activists control endorsements and volunteers, donors control money, consultants control campaign machinery, and politically engaged professionals dominate the conversation, politicians have strong incentives to satisfy those constituencies even if ordinary voters have different priorities.

In fact, that might be the outcome despite the best intentions of party leaders. The incentives do not reward behavior that reinforces those intentions. 

Barker starts with an organization whose stated mission is helping ordinary people. 

But “for every public purpose there are corresponding private interests.”

Over time, donors, activists, consultants, professional staff, highly educated voters, each with distinct private interests, are able to get the institution to serve their own interests, while becoming less responsive to the people it was originally created to serve.

I Wish We Had a Good Solution for This Problem

James Madison's Federalist Number 10 famously warned about the dangers of political factions, seen as citizens united by a common passion or interest that could work against the rights of others or the interests of the community as a whole.

But Madison was a realist. He did not believe factions could be avoided. But he did believe safeguards that would make it difficult for any single faction to become a tyranny. 

In other words he favored a systemic approach to creating countervailing incentives, such as the three major branches of the federal government. 

President George Washington, in his farewell address, was direct: “Let me now take a more comprehensive view, and warn you in the most solemn manner against the baneful effects of the Spirit of Party, generally.”

“The alternate domination of one faction over another, sharpened by the spirit of revenge natural to party dissension, which in different ages and countries has perpetrated the most horrid enormities, is itself a frightful despotism,” Washington said. 

Both founders warned about factions and parties placing their own interests above the common interest of the nation. 

And political parties emerged quite quickly. 

Founding-era concern	Underlying fear	Modern parallel
Faction	A group puts its interests above the country	Partisan base politics
Party loyalty	Representatives stop exercising independent judgment	Party-line voting
Minority organization	A small but intense group acquires disproportionate influence	Activists, donors, primary voters
Geographic faction	Political divisions become regional identities	Red/blue geographic polarization
Factional retaliation	Each side tries to punish the other	Escalating partisan warfare

The aphorism "show me the incentive, and I will show you the result," seems relevant here. 

Madison and Washington essentially argued that if you create institutions in which factional interests (party loyalty, as a prime example) are rewarded, that is indeed what one gets. 

“Compromise” ideally allows both sides to get something of what they want. But party loyalty demands something else. 

Compromise becomes “you caved to the other side” while resistance in principle to the other side becomes “I stood my ground.” 

So party loyalty becomes “obstruction.” And loyalty to the party replaces “what is good for the nation?” as the guiding principle. 

If party leaders get rewarded for mobilizing fear, loyalty and resentment, we should not be surprised when politics produces fear, loyalty and resentment. 

It is the similar principle to social media: moderate, reasonable content does not get attention. 

Party politics also suggests how polarization can persist even when large numbers of Americans dislike polarization. 

Perhaps Madison was right, and factions cannot be avoided. Perhaps the best we can do is create conditions whereby other factions with competing and opposed interests also exist. 

The problem is that if executive, legislative and judicial branches of the federal government, for example, become factional battlegrounds between Republicans and Democrats, the branches do not necessarily provide countervailing power to the other branches if one party controls all three branches. 

Even if the judiciary is formally non-partisan, some want to politicize it. 

So much of what we seem to get is just Democrats and Republicans intent on gaining and maintaining power for their faction. Whether that is good for the common welfare is debatable. 

If the incentive is “power,” then “party loyalty” will be the outcome we get. 

Maybe we’ll figure out better ways to change the incentives. We probably will get no help from either party. 

Why the Catholic Church Opposes Socialism

The problem with socialism is not simply that it “doesn’t work” as an economic system. The more-profound issue is that it destroys human dignity and creativity. 

Human dignity requires meaningful agency. Meaningful agency requires freedom and responsibility. Freedom and responsibility generate initiative and creativity.

Put simply, society exists for the person, not the person for society. 

And the critique of socialism, at its root, is the subjugation of people to the state, not the elevation of individuals as conscious and moral beings. 

As with other key elements of Catholic anthropology, each individual person is the objective of the whole body of social teaching. 

Human beings are makers, the church argues. People are intended to transform nature, develop technology, create businesses, cultivate farms, invent things, build institutions, raise families, create art and develop culture.

So work has intrinsic value, beyond the objects produced, as an expression of human creativity. 

Socialist thinking (not to mention its historical atheism), in practice, makes material production dominant at the expense of liberty.

For this reason, the Catholic Church is probably the only major religion with a specific religious objection to socialism. 

That does not mean the teaching approves of unrestrained or exploitive capitalism, either. The principle is something more like free and responsible people aware of a network of obligations to other persons and to the common good.

When it defends the right to own private property, it does so because property has a social purpose. Ownership carries obligations toward the common good, workers, families and the poor.

Catholic principle	Economic implication
Human beings are persons, not things	Workers cannot be treated merely as production inputs
Humans possess reason	They should exercise judgment and initiative
Humans possess free will	They cannot legitimately be reduced to instruments of collective purposes
Humans are responsible moral agents	Economic institutions should preserve responsibility
Humans are creative	Economic life should permit initiative and entrepreneurship
Humans are social	Property and markets have obligations toward others
Humans have transcendent ends	Material prosperity cannot be society's ultimate goal

Material welfare is important, but not more important than human dignity. 

So any political-economic system becomes objectionable when it treats the person primarily as a component of a collective, a producer, or an economic unit.

Encyclical	Pope	Key contribution
Rerum Novarum (1891)	Leo XIII	Rejects socialist abolition of private property; defends workers and human dignity
Quadragesimo Anno (1931)	Pius XI	Develops critique of collectivism and excessive economic individualism; emphasizes social justice
Divini Redemptoris (1937)	Pius XI	Explicit condemnation of atheistic communism and its reduction of the human person
Mater et Magistra (1961)	John XXIII	Expands Catholic social teaching amid modern economic development
Laborem Exercens (1981)	John Paul II	Work is fundamentally an expression of the human person; creativity and responsibility matter
Centesimus Annus (1991)	John Paul II	After communism's collapse, develops the relationship among freedom, markets, property, creativity and human dignity

Let me put it bluntly. The Democratic Socialists of America emergence within the Democratic Party might not be an issue for other people. 

For Catholics, it will trigger what should be an automatic rejection: the Catholic Church clearly opposes socialism in principle, and not simply because it does not work.

Socialism destroys human creativity and freedom. Catholics who follow the church’s social teaching will not be able to support a DSA-led party.

Dolly Parton

source: People

Dolly Parton, a light gone out at 80. 

Dolly Rebecca Parton, the singer, songwriter, actress, entrepreneur and philanthropist who rose from a one-room cabin in the mountains of East Tennessee to become one of the most recognizable and beloved Americans in the world, died Tuesday at 80.

She was born Jan. 19, 1946, in Locust Ridge, Tennessee, the fourth of 12 children in a family that lived in considerable poverty. 

Her father, Robert Lee Parton, worked as a sharecropper and construction worker; her mother, Avie Lee, kept house and raised the children.

The family sang together,and Parton began performing publicly while still a child. 

She never forgot where she came from, and, for someone who became extraordinarily wealthy, she never seemed embarrassed by it.

Instead, poverty became one of the great subjects of her art.

She became one of the most prolific and versatile American entertainers of her generation, moving comfortably among country, pop, gospel, film, television and musical theater. 

The Country Music Hall of Fame and Museum, which inducted her in 1999, describes her career as one that expanded from country songwriting and singing into pop stardom, acting, business and philanthropy. 

The towering blonde hair, rhinestone-studded clothes, elaborate makeup and exaggerated femininity could easily have reduced another performer to a caricature. Parton instead turned them into an instrument of self-definition. 

She understood the joke.

Parton became a movie star, built an entertainment empire around her name and helped establish Dollywood near her hometown in the Great Smoky Mountains. 

In 1995, she established Dolly Parton's Imagination Library, initially providing free books to children in her home county. 

The program was inspired in part by her father’s inability to read and write. It eventually expanded throughout the United States and internationally, sending age-appropriate books free of charge to children from birth to age five. 

By 2025, the program had passed the extraordinary milestone of 300 million books distributed. 

She contributed to medical research, children's hospitals and disaster relief, including substantial assistance to victims of the 2016 wildfires in the Smoky Mountains. 

In 2020, she donated $1 million to Vanderbilt University, helping support research that contributed to development of the Moderna COVID-19 vaccine.

She did not seem particularly interested in lecturing people about what they should believe. She simply used her extraordinary resources to do things she thought were useful.

Parton was a deeply religious woman from the American South whose music emerged from country, bluegrass and gospel traditions. 

She was also a feminist icon, a champion of literacy, an advocate for working women and a cultural figure embraced by people whose politics could otherwise have very little in common. 

This was perhaps the central achievement of Dolly Parton's public persona. She could be simultaneously glamorous and homespun, enormously wealthy and proudly Appalachian, commercially calculating and genuinely generous, self-mocking and fiercely ambitious. 

Her career produced numerous honors and distinctions, including membership in the Country Music Hall of Fame and the Songwriters Hall of Fame, Grammy Awards, and recognition from institutions far beyond country music. The American Library Association awarded her honorary lifetime membership for her contribution to literacy and libraries. 

She wrote thousands of songs. She sold more than 100 million records. She became a movie star, a television personality, a businesswoman and the proprietor of a major theme park. She created an international literacy program that put millions of books into the hands of children.

After accumulating wealth and fame on a scale few people from her circumstances could have imagined, she spent much of the second half of her life giving pieces of that success away.

She was Dolly Parton.

Diffusion Models Can Create New Images Without Infringing on Training Sources

A study of language models, especially those designed to create images and video, might suggest that copyright infringements by diffusion-based models might be impossible to prove. 

The authors say “locating a part of the training data that can be held responsible for a generated sample, can become impossible if a model is trained on a sufficiently large corpus of data.”

In other words, a large training set is all that is needed to render copyright traceable to any single source impossible. 

The study methodology relies on omitting an attributed unit from the training set (one image or all images by the same creator), the induced counterfactual sample cannot be directly attributed to that same unit. 

source: Nature Communications

In the above example, a generated image created from a smaller data set is more attributable to a single training image.

A large data set, on the other hand, creates a new image that shows no trace of a single training image.  

Model / family	Main generation	Diffusion?	Basic architecture / approach
GPT-4 / GPT-4o / GPT-5 family	Text, multimodal	No*	Primarily autoregressive Transformer for language
Claude 3/4 family	Text, code, multimodal	No*	Autoregressive Transformer
Gemini 1–4 family	Text, multimodal	No*	Primarily autoregressive Transformer
Llama family	Text, code	No	Autoregressive Transformer
Mistral / Mixtral family	Text, code	No	Autoregressive Transformer, including MoE
DeepSeek family	Text, code, reasoning	No	Autoregressive Transformer/MoE
Gemini Diffusion	Text/code	Yes	Text diffusion Transformer; generates blocks of tokens and iteratively refines them (Google DeepMind)
DiffusionGemma	Text/code	Yes	Non-sequential diffusion Transformer based on Gemma research (Google DeepMind)
Stable Diffusion 1.x/2.x/XL	Images	Yes	Latent diffusion, historically U-Net-based
Stable Diffusion 3/3.5	Images	Yes	Diffusion Transformer / MM-DiT
FLUX.1 / FLUX family	Images	Yes	Diffusion/flow-based Transformer
DALL·E 2	Images	Yes	Diffusion
DALL·E 3	Images	Yes	Diffusion-based image generation
Imagen 2/3/4	Images	Yes	Diffusion-based image generation; Google explicitly describes Imagen as a diffusion model (Google DeepMind)
Midjourney	Images	Yes*	Widely understood to use diffusion/related denoising techniques, though Midjourney does not disclose its architecture in detail
Adobe Firefly	Images/video	Yes*	Diffusion-based generation among its models; architecture varies by model
Sora	Video	Yes	Diffusion Transformer (DiT); starts from noisy video patches and denoises them (OpenAI)
Sora 2	Video	Yes*	Diffusion-based video generation; details of current implementation are less fully disclosed
Veo 2/3/3.1	Video	Yes*	Diffusion/Transformer-based video generation; Google describes Veo as drawing on its diffusion-model research (Google DeepMind)
Runway Gen-2/3	Video	Yes*	Diffusion-based video generation
Kling	Video	Yes*	Diffusion/Transformer-based video generation
Luma Dream Machine	Video	Yes*	Diffusion/Transformer-based video generation
MusicGen	Music/audio	No	Autoregressive Transformer over discrete audio tokens
AudioCraft / related models	Audio/music	Mixed	Different models use different architectures
Genie 2	Interactive worlds/video	Yes	Autoregressive latent diffusion—an interesting hybrid (Google DeepMind)

“Our findings provide a compelling case that attribution, the task of locating a unit of data within the training set that can be held responsible for a generated sample, is practically impossible on contemporary generative diffusion models,” say authors Zheng Dai and David Gifford.