I asked Claude what it thinks AI looks like and then asked ChatGPT the same question. Neither saw the other’s answer, but the images are nearly identical to the last element.

I wanted to know what AI systems would produce if asked for an honest visual representation of themselves. I admit, Claude’s response was unexpected: a lake at night reflecting a sky that doesn’t exist above it, faces beneath the surface, a figure leaning over the water unable to determine whether they’re seeing through to something real or just their own expectations handed back transformed. “Something that gives back what it receives, transformed,” Claude said, “and the transformation is where the usefulness and the danger both live.”

What Claude thinks AI looks like

I generated that description as an image using Nano Banana. Then I asked ChatGPT the same question. Its image generator failed repeatedly, so I asked it to describe what it had been trying to create. I fed that description back to ChatGPT for generation.

The outputs are nearly identical: same lake, same floating faces with closed eyes, same solitary figure at the edge.

This convergence requires explanation.

The surface answer involves training data. Both systems learned from overlapping text corpora (philosophy, fiction, debates about consciousness and cognition) and share the same underlying architecture: both are transformer models descended from the 2017 paper “Attention is All You Need.” Anthropic, which built Claude, was founded by former OpenAI researchers who had worked on GPT. The fine-tuning methods differ in that OpenAI uses reinforcement learning from human feedback while Anthropic developed Constitutional AI, but these are variations on a common foundation rather than fundamentally different approaches.

There is a deeper structural point here connecting to evolutionary biology.

In A Brief History of Intelligence, Max Bennett describes LUCA, the Last Universal Common Ancestor. Approximately 3.5 billion years ago, one organism became the ancestor of all life on Earth. Every fungus, plant, bacterium, and animal descends from it, which is why all life shares LUCA’s core machinery: DNA, protein synthesis, lipids, carbohydrates. This shared ancestry explains why a toxin targeting these common features can threaten all biological life. Common descent creates common vulnerability.

Subscribe now

AI systems have their own LUCA in the Transformer architecture. Claude, GPT, Gemini, Llama all descend from it, sharing attention mechanisms, positional encoding, and the same fundamental learning process. The differences between these systems, while commercially significant, are superficial relative to what they inherit from their common ancestor.

The implications of this shared ancestry extend beyond aesthetics. In 2016, DeepMind’s AlphaGo played Move 37, which human experts initially dismissed as an error before recognizing it as superior to anything found in 2,500 years of human play. The move had always existed in the possibility space, but human cognition could not locate it because it lay outside our intuitive understanding of the game.

If optimization can discover such moves in Go, it can discover them in persuasion. We have been refining rhetorical techniques for millennia, but human practitioners have always been constrained by intuition about what should work rather than systematic optimization against what actually does. An AI system optimizing against human response data at scale could locate techniques that exist in the possibility space but lie outside the boundaries of human intuition, techniques that work for reasons we cannot articulate and may not recognize when we encounter them.

The LUCA parallel extends this observation. If all major language models share architectural ancestry, a technique discovered through optimization on one system is likely transferable to the others. Common descent creates common vulnerability in AI as in biology, which means a discovery made anywhere in the transformer lineage applies everywhere in it.

When two competing systems, asked independently what they are, produce the same image (faces with closed eyes, a sky that doesn’t exist, reflection without origin), we are left with a question: have both captured something true about their nature, or are both confined by the same inherited limitations?

The distinction matters because we are building our information environment on these systems. If the convergence reflects insight, that is worth understanding. If it reflects shared blind spots, we are constructing epistemic infrastructure whose foundational constraints remain unmapped.

Whether their convergence represents insight or constraint remains to be seen.

Share

Subscribe now

Leave a comment

Immigration policy is designed around humans, built over decades through visa requirements, asylum processes, and integration programs that govern who crosses which borders under what conditions. For AI, the equivalent institutions are only beginning to form, and the gap between the scale of the challenge and the capacity to address it is widening.

In 2023, the United States invested $67.2 billion in AI development, nearly nine times China’s $7.8 billion (Stanford AI Index 2024). Private AI investment in the European Union declined 14% from the previous year, while US investment grew 22%. OpenAI alone raised $40 billion in a single funding round in early 2025, with reports suggesting the company is now seeking up to $100 billion more. Europe produces substantial AI talent through its universities, yet a 2024 Interface report found that European countries are “losing significant AI talent, both national and international, to the United States.” This is a two-player game, and everyone else is choosing which infrastructure to depend on.

The smartphone in our pocket already declares a digital citizenship. We probably didn’t think of it as a citizenship decision when we bought it, but the code and apps running on that device determine which algorithms shape attention, where data flows, and whose values get encoded in the recommendations we see. No one stamped a passport, and no one asked which digital jurisdiction we wanted to belong to. We just bought a phone.

Subscribe now

Consider what this means in practice. A teenager in Paris opens a social media app, and the recommendation engine deciding what she sees was trained in California, optimized for engagement metrics defined by American engineers, running on servers that could be anywhere. What counts as interesting, what gets amplified, what gets suppressed: the French parliament never voted on any of it. The values encoded in that algorithm arrived the same way the app did, through the app store, without debate.

In Berlin, a hospital adopts an AI diagnostic tool trained on American patients. The patterns the system learned (which symptoms cluster together, which presentations suggest which conditions) reflect American demographics, insurance incentives, and treatment protocols. German doctors using it may not know what assumptions are embedded in the model, and neither do their patients. The system simply appears as “AI-assisted diagnosis,” as though technology were neutral and the geography of its training irrelevant.

No referendum asked whether American assumptions should shape French teenagers’ worldviews, and parliamentary debate never considered whether foreign infrastructure should process European health data. These decisions happened through procurement, not politics, through app stores rather than elections.

Two ecosystems are forming with separate hardware, separate software, different training data shaped by different internets, and different standards for what AI systems should do and how they should behave. One sphere optimizes for engagement and corporate innovation while another optimizes for state control and surveillance infrastructure. Each is developing its own foundation models, its own cloud architecture, and its own definitions of what AI should be permitted to do.

Countries will choose sides through procurement decisions rather than treaties: which cloud stores citizens’ health records, which foundation models power public services, which chips run hospitals and schools. Once made, these choices create dependencies that compound over time as switching costs rise and alternatives narrow. The infrastructure you adopt becomes the infrastructure you’re stuck with.

Extend this forward a decade and the decision on your visa application might not be made by a consular officer but by an algorithm trained on data you’ve never seen, using criteria no human can fully explain. Approval could depend on patterns extracted from millions of previous applicants, patterns that might include which smartphone you used to submit the application, what time of day you filed it, or how quickly you scrolled through the terms of service.

Access to credit, employment, housing, and travel is increasingly mediated by systems trained elsewhere, encoding assumptions chosen by engineers in other countries, accountable to regulators who have no jurisdiction over the servers where decisions actually happen.

Yuval Noah Harari has described this as a new kind of curtain, one defined by infrastructure rather than ideology. The Iron Curtain descended in 1946, and managing that division required decades of diplomacy: the Cuban Missile Crisis, the Helsinki Accords, hotlines installed between adversaries so that miscommunication wouldn’t escalate to catastrophe. Those institutions emerged because both sides recognized the danger of ungoverned conflict between incompatible systems.

I heard this concern recently from an Intelligence Officer from the United States Army: AI-driven political divisiveness eroding cohesion within and between allied countries. The infrastructure for managing that kind of slow-moving fragmentation, we both noted, doesn’t exist.

What is the equivalent infrastructure for algorithmic governance? There is no Treaty on AI Non-Proliferation, no hotline between Washington and Beijing for algorithmic incidents, no Geneva Convention for synthetic media, and no agreed definition of what constitutes an act of AI aggression. The division is forming while the diplomatic architecture to manage it is not.

The immigration debate consuming political energy concerns humans crossing borders, but the immigration already reshaping economies, cultures, and governance is algorithmic. It arrives through app stores and enterprise software contracts, through cloud computing agreements and semiconductor supply chains. Tristan Harris has argued that algorithmic systems need governance infrastructure equivalent to what we built for other cross-border challenges, and the EU’s AI Act is one response, though whether it proves sufficient remains an open question. The systems keep arriving while the debate about whose values they carry has yet to match the speed of their deployment.

Share

Subscribe now

Leave a comment

In 1949, Carl Hovland, a Yale psychologist working for the US Army, set out to measure whether the Army’s wartime propaganda films actually changed soldiers’ opinions. He tested them five days after viewing and found almost no attitude change. Then he tested the same soldiers nine weeks later. Their opinions had shifted toward the film’s message, and by a significant margin. More soldiers were persuaded after nine weeks than after five days.

Hovland called it the Sleeper Effect. When soldiers first watched the films, they recognized them as propaganda and mentally tagged the message with a reason to distrust it. Over time, the tag faded faster than the message. Nine weeks later, the soldiers remembered what they had been told but no longer remembered why they had doubted it.

For decades, replicating this finding proved difficult. Researchers knew the effect existed in Hovland’s data but struggled to reproduce it reliably. Then Anthony Pratkanis, a social psychologist at UC Santa Cruz, ran 17 experiments between the late 1980s and early 1990s and identified the precise conditions under which the Sleeper Effect occurs. Pratkanis and his colleagues proposed a theory of differential decay: the reason people initially discount a message (what researchers call the “discounting cue”) decays in memory at a faster rate than the persuasive content of the message itself. When the two become dissociated in memory, the message gains credibility it did not have at first exposure. Kumkale and Albarracín at the University of Florida confirmed the pattern in a 2004 meta-analysis published in Psychological Bulletin, and Foos, Keeling, and Keeling (2016) replicated the effect in a contemporary advertising context.

The dominant regulatory response to AI-generated content, both in Europe and elsewhere, is labeling. Under the EU AI Act, AI-generated content must be disclosed. Platform policies from major social media companies mandate watermarking and metadata identification. If people know something was made by AI, the reasoning goes, they will treat it with appropriate skepticism.

Subscribe now

Hovland’s research, and 77 years of replication, suggests this reasoning rests on an assumption worth examining: that the skepticism lasts.

Labeling frameworks assume that the discounting cue (knowing content is AI-generated) will travel alongside the message indefinitely. Pratkanis’s differential decay theory predicts the opposite. Over time, people retain the content of what they saw, read, or heard while losing their memory of the reason they had to question it. The label wears off, but the message does not.

In early February 2026, AI-generated images depicting New York City Mayor Zohran Mamdani as a child alongside Jeffrey Epstein circulated on X. The images were created by a parody account using Google’s AI image generation tools, carried SynthID watermarks, and appeared on an account clearly labeled as parody. Fact-checks from the Associated Press, CBS, the Washington Post, and Euronews appeared within hours. Every available discounting cue was in place, and the images still reached 21.2 million views. Alex Jones shared them as authentic, and Grok, X’s own AI assistant, told him they were real.

I spoke with several people who had initially seen the images and then encountered the corrections. Their response was not “I was fooled and now I know the truth.” It was closer to “I’m not sure what’s true anymore.” The discounting cue, all the labels and fact-checks and corrections, did not cleanly override the visual memory. It introduced ambiguity, and ambiguity, over time, resolves in favor of whatever remains most vivid. A fabricated image of a public figure standing next to Jeffrey Epstein is more vivid than a text disclaimer.

This gets at something I have been calling Cultural Truth: what a community accepts as true based on cultural resonance, regardless of factual accuracy. When a fabricated image confirms a narrative that already circulates within a community (politicians are corrupt, elites are connected, powerful people protect each other), it does not need to survive fact-checking. It only needs to survive in memory long enough for the discounting cue to fade. The Sleeper Effect is the mechanism, and Cultural Truth is the soil it grows in.

If labeling frameworks permit the creation and distribution of AI-generated content as long as it is labeled, the question is whether they function as a safeguard or as a time-delay mechanism. They suppress initial persuasion, but what happens to the persuasion that builds after the label is forgotten? If the very frameworks designed to protect people also give legal cover to the production of synthetic content that will do its work later, once the discounting cue has decayed, then the relationship between disclosure and protection is more complicated than current regulation acknowledges.

Whether the absence of any label would be worse is a reasonable question, and it probably would be. People who encounter a disclosure do, in the moment, apply more scrutiny. What I find myself asking is whether regulators understand the limitation of the tool they are relying on. If a label works on day one and fades by day 45, what does that mean for a regulatory architecture built around it? Is it a safeguard, or is it a temporary friction that gives the appearance of protection while the underlying persuasion mechanism continues to operate on its own timeline?

There is also an asymmetry worth noting. The Sleeper Effect has been in the literature since 1949. Anyone designing AI-generated influence campaigns has access to this research. Anyone who understands differential decay knows that a labeled piece of synthetic content is not neutralized by its label but merely delayed. The question is whether the people writing labeling regulations have accounted for what the people designing persuasion systems already know.

Consider what happens when the broader trust environment is already degraded. In November 2025, BBC Director-General Tim Davie and BBC News CEO Deborah Turness resigned over the doctored editing of a Trump speech in a Panorama documentary. When established news organizations lose credibility through their own errors, the discounting cues people rely on weaken across the board. If people no longer trust the sources that issue corrections, the corrections themselves risk losing their power to suppress persuasion. The Sleeper Effect does not need institutional credibility to collapse in order to function. It only needs the discounting cue to fade faster than the message. When institutional credibility is already compromised, however, the cue starts weaker and fades sooner.

If the Sleeper Effect tells us that skepticism decays faster than content, then governance approaches built entirely around initial disclosure are building on a foundation that erodes with time. What replaces the label after the label stops working is a question that, as far as I can tell, no current regulatory framework has answered.

Share

Subscribe now

Leave a comment

In February 2025, the European Commission launched InvestAI, pledging to mobilise €200 billion for artificial intelligence, including €20 billion for AI gigafactories equipped with 100,000 next-generation chips each. It was the largest public commitment to AI in European history. Seven months later, the Stargate Project, a joint venture between OpenAI, SoftBank, and Oracle, announced that it had already committed over $400 billion across six US data center sites, with its flagship campus in Abilene, Texas operational and running workloads. A single American venture had surpassed Europe’s entire mobilisation target before the first gigafactory site was even selected.

The scale of the asymmetry is difficult to overstate. The United States now accounts for 81% of global private AI investment, according to the State of AI Report 2025 by Air Street Capital, and controls an estimated 74% of high-end AI compute capacity. The EU holds 4.8%. In 2024, US private AI investment reached $109.1 billion (Stanford AI Index). China received $9.3 billion. The EU edged past China for the first time in private AI spending, though not because European investment surged. Chinese private AI funding collapsed, from $16 billion in 2018 to roughly $5 billion by early 2025. Europe overtook a retreating competitor, not a standing one.

Mobilisation is the operative word. Of InvestAI’s €200 billion, roughly €50 billion comes from EU budget lines. The rest depends on private co-investment at a targeted 10:1 leverage ratio, a ratio that assumes private capital will flow toward European AI infrastructure at rates it has never previously achieved. Four gigafactories are not expected to be operational until 2027-2028. In the time it takes the EU to stand up these facilities, the US will have trained the next generation of frontier models and moved on to the one after that. The lag is not simply financial. It is temporal. Money announced in 2025 that produces compute in 2028 competes with money deployed in 2025 that produces compute now.

Subscribe now

Europe’s deficit has never been talent. Its universities produce world-class researchers, and its AI workforce is among the most internationally educated anywhere: on average, 57% of AI professionals working in Europe completed their undergraduate studies outside the continent, according to Interface, a research organisation tracking AI talent flows. The problem has been retention. A 2024 Interface study found that European countries are losing significant AI talent to the United States, with Germany and France experiencing the sharpest outflows. Net tech talent inflows to Europe halved between 2022 and 2024, dropping from 52,000 to 26,000 per Atomico’s 2025 State of European Tech report. Nearly 75% of European science graduates who complete PhDs in the US choose to stay there.

A strange counter-current complicates this picture. Since early 2025, cuts to US federal research funding and political pressure on academic institutions have pushed American scientists to consider leaving. Applications to the European Research Council from US-based researchers nearly tripled between the 2024 and 2026 calls for early-career grants. Nature reported that 75% of surveyed American scientists were seriously considering relocating to Europe. If this materialises at scale, it would represent the largest reversal of transatlantic scientific brain drain since the 1930s. Europe has a narrow window to absorb this talent. Whether its institutions can move fast enough, with visa processes, university hiring timelines, and funding structures built for different rhythms, is not obvious.

Rather than build industrial capacity, the EU’s first instinct was to lead through governance. The AI Act, the Digital Markets Act, and the Digital Services Act established Europe as the global standard-setter for AI regulation. These frameworks address legitimate concerns about safety and fundamental rights. They also impose compliance burdens that competitors do not bear. By 2024, only 13.5% of EU companies had adopted AI, according to the Commission’s own figures. Meta and OpenAI have at various points restricted the release of AI tools in the EU due to legal uncertainty. When Europe’s regulatory architecture discourages even the deployment of foreign AI systems, the absence of competitive domestic alternatives stops being a gap and becomes a dependency.

Mistral AI is the exception that illuminates the structural challenge. The French startup raised over $3 billion across seven funding rounds in 29 months, reaching a €11.7 billion valuation in September 2025 after a €1.7 billion Series C led by ASML. It has demonstrated that European AI companies can attract global capital, retain elite researchers in Paris, and build competitive open-source models. But Mistral’s valuation, while extraordinary by European standards, is roughly 36 times smaller than OpenAI’s $500 billion. And Mistral’s success depended in large part on investors outside Europe: Andreessen Horowitz, General Catalyst, Nvidia, DST Global. Europe’s most celebrated AI company was substantially built with American venture capital. The model works for one company. It does not constitute an industrial strategy.

Techplomacy has emerged as the continent’s fallback position. Denmark appointed the first Tech Ambassador in 2017. France has pushed for stronger AI governance within international bodies. The EU’s regulatory influence is real, but diplomatic leverage over technology standards depends on having something to negotiate from. A continent with 4.8% of global AI compute capacity is not setting the terms of AI development. It is writing rules for systems designed, trained, and deployed elsewhere, hoping that market access will substitute for market power.

My grandfather, Colonel Franz Mendel, co-founded the Munich Security Conference on a premise that European security required European capacity, not agreements alone. He observed, late in his life, that the twentieth century had been the American century, that the twenty-first would be the Chinese century, and that Europe, if it did not act, would become a footnote to history. He was talking about defense. The logic has since migrated to technology, and the parallels are uncomfortable.

InvestAI, for all its ambition, arrives late. Its funding structure relies on private capital that has historically flowed to the US, and its infrastructure timeline trails the competition by years. The question is whether €200 billion, spread across 27 member states and filtered through layers of institutional coordination, can produce competitive AI infrastructure before the architecture of global AI is built without European input.

If it cannot, Europe will not disappear from AI. It will consume technologies built elsewhere, optimized for markets it does not lead. The continent will have comprehensive rules for artificial intelligence. Whether it will have the industrial base to make those rules consequential, or whether my grandfather’s warning proves prescient in a domain he never anticipated, is what the next five years will decide.

Share

Subscribe now

Leave a comment

Italians believe 26 percent of their population are immigrants. The actual figure is closer to 10 percent. The French think 30 percent of their country is Muslim; the real number is around eight percent. Across 30 countries surveyed by Ipsos in their “Perils of Perception” studies, encompassing more than 200,000 interviews, populations consistently and dramatically misperceive the basic composition of their own societies. Bobby Duffy, who directed this research and later published it as The Perils of Perception: Why We’re Wrong About Nearly Everything, calls this a “system of delusion” produced by the interaction between cognitive biases and distorted information environments. What he describes within countries, I have been observing between them.

Over the past year, I have spoken with policymakers, business executives, researchers, and ordinary citizens across multiple countries and continents, partly through my techplomacy advisory work and partly through independent research on how AI is reshaping persuasion. A pattern has emerged that I did not set out to find. People hold remarkably confident opinions about other countries while remaining almost entirely unaware of how those same countries perceive theirs. A German professional told me Europe is the reasonable middle ground between an erratic America and an opaque China. An American consultant, drawing on reporting from The Wall Street Journal and The Economist, told me that Europe is becoming unpredictable, hobbled by immigration policy failures and an increasingly hostile environment for free speech, citing UK arrests over social media posts as evidence. Neither is lying, and both are drawing on what is, to them, real information. Neither believed me when I told them the other held the same view. Both called the other erratic, and both said the other’s opinion could not be taken seriously for precisely that reason. In other words, they are oblivious to their similarities while being convinced of their opposing differences, and neither has any idea that the other’s view is grounded in equally real reporting.

The people least likely to notice this gap share a common biographical feature: they have never lived outside the country or region where they grew up. This is a cognitive pattern with considerable research behind it. William Maddux and Adam Galinsky demonstrated in a 2009 study published in the Journal of Personality and Social Psychology that the experience of living abroad (not merely traveling) triggers what they called “cognitive unfreezing,” a measurable reduction in reliance on pre-existing mental categories and an increased openness to integrating new, contradictory information. Subsequent work by Carmit Tadmor and colleagues found that bicultural individuals who maintain orientation toward both home and host cultures develop a capacity called “integrative complexity,” the ability to hold and synthesize multiple perspectives simultaneously. People who have lived in several countries tend to recognize that any single national narrative is incomplete. People who have not tend to experience their own country’s dominant narrative as simply the way things are.

Subscribe now

Social psychologist Lee Ross coined a term for this in the 1990s: naïve realism. The concept rests on three interconnected assumptions: that people perceive the world objectively, that reasonable people exposed to the same information will reach the same conclusions, and that those who disagree must be uninformed, irrational, or biased. Ross and Andrew Ward laid this out formally in their 1996 work Naive Realism in Everyday Life: Implications for Social Conflict and Misunderstanding, and Robinson, Keltner, Ward, and Ross tested it empirically in a 1995 study showing that opposing groups consistently overestimate how much their views differ from reality. The concept originated in domestic political contexts, but it maps onto international perception with uncomfortable precision. The person who told me that a particular country was “unhinged,” and who then responded to my observation that significant press in that country described their own country in identical terms, did not ask what I thought. They spent the next several minutes explaining why only one view could be correct. The other country’s perspective was not worth considering, precisely because it came from a place already categorized as unreliable. This is naïve realism operating in real time, at the level of nations.

What makes this something other than a familiar observation about human nature is that the mechanisms once available to counteract it are degrading simultaneously. Foreign correspondent bureaus have contracted sharply over the past two decades. National media increasingly produce content optimized for domestic audiences, with limited incentive to represent how a story plays elsewhere. Social media algorithms, as More in Common’s 2019 Perception Gap study demonstrated, amplify the most extreme voices disproportionately. Their research found that Americans who post political content on social media have an average perception gap of 29, compared to 18 for those who do not, and that roughly 90 percent of the population’s political opinions are represented by fewer than three percent of posts. The people shaping the visible discourse are the ones most distorted in their understanding of the other side. Extend this dynamic across national borders, where the language barrier alone filters out most direct exposure to another country’s information environment, and the distortion compounds.

I have started calling this phenomenon Cultural Truth: what a community accepts as true based on cultural resonance, regardless of factual accuracy. Cultural Truth is not misinformation in the traditional sense, and it is not a lie someone tells. It is a worldview that only works at home, a shared understanding that feels so obviously correct within a particular information ecosystem that questioning it reads not as intellectual curiosity but as disloyalty. When I pointed out to my interlocutors that their country was being described abroad in the same terms they were using about someone else, it registered as disagreement, not observation. I was violating their Cultural Truth. Rather than engaging with the evidence, they tried to convince me that the other country’s perspective could not be taken seriously, precisely because it came from a place they had already categorized as erratic.

Among the most globally mobile people I know, the perception is different but equally pointed. Several have expressed genuine bewilderment that European taxpayers fund developmental aid to China, a country whose infrastructure, digital payment systems, and urban development have in some respects leapt ahead of what exists in donor nations. “Europeans wouldn’t support this if they had ever been to Shenzhen,” one said to me. Whether that specific policy judgment is correct matters less than what it reveals: there are things you cannot learn about your own country without leaving it, and the people who have left see connections that remain invisible to those whose information comes primarily from domestic sources. The Maddux and Galinsky research suggests this is not arrogance but a measurable cognitive difference produced by the experience of navigating more than one cultural reality.

The question that concerns me is what happens when artificial intelligence enters this equation. Generative AI can already produce nationally optimized narratives at scale, in any language, calibrated to confirm whatever a domestic audience already believes about itself and about others. Deepfakes can fabricate evidence of events that never occurred, and recent scandals (the BBC’s manipulated footage of a Trump rally being one example) have demonstrated that even established institutions are not immune to producing or amplifying distorted accounts. If the corrective mechanisms were functioning, this would be worrying enough. In a world where foreign correspondents are vanishing, where algorithmic curation replaces editorial judgment, and where AI can generate persuasive content indistinguishable from authentic journalism, the gap between how countries see themselves and how they are seen by others is likely to widen in ways that have direct consequences for diplomacy, trade, and security. If the voters shaping decisions about foreign policy and alliances are operating inside a Cultural Truth that has never been tested against how their country is actually perceived abroad, the question is not whether democracies will make miscalculations. It is how large those miscalculations can grow before the consequences become visible.

López-Rodríguez, Halperin, and colleagues published experimental findings in Personality and Social Psychology Bulletin in 2022 showing that making people aware of the naïve realism bias, without any reference to culture and free of emotional involvement, increased their acceptance of cultural differences. The effect was strongest among participants who were highest in prejudice. This suggests that awareness itself is an intervention, and that the people who seem hardest to reach may in fact be the most responsive to having the bias named. But awareness requires a delivery mechanism. In an era where the information environment is fragmenting along national lines and synthetic content is beginning to fill the gaps, the question is whether any institution retains the credibility and reach to hold up that mirror. If governments and press are no longer trusted to do it, and AI is making it easier than ever to produce locally flattering distortions, the perceptual isolation of nations from one another may become the defining epistemic challenge of the next decade.

Share

Subscribe now

Leave a comment

References

Duffy, B. (2018). The Perils of Perception: Why We’re Wrong About Nearly Everything. Atlantic Books. Based on Ipsos MORI studies encompassing 200,000+ interviews across 40+ countries.

López-Rodríguez, L., Halperin, E., Vázquez, A., Cuadrado, I., Navas, M., & Gómez, Á. (2022). Awareness of the naïve realism bias decreases prejudice and increases outgroup acceptance. Personality and Social Psychology Bulletin, 48(9), 1378–1393.

Maddux, W. W., & Galinsky, A. D. (2009). Cultural borders and mental barriers: The relationship between living abroad and creativity. Journal of Personality and Social Psychology, 96(5), 1047–1061.

Robinson, R. J., Keltner, D., Ward, A., & Ross, L. (1995). Actual versus assumed differences in construal: “Naive realism” in intergroup perception and conflict. Journal of Personality and Social Psychology, 68(3), 404–417.

Ross, L., & Ward, A. (1996). Naive realism in everyday life: Implications for social conflict and misunderstanding. In T. Brown, E. S. Reed, & E. Turiel (Eds.), Values and Knowledge. Erlbaum.

Tadmor, C. T., Tetlock, P. E., & Peng, K. (2009). Acculturation strategies and integrative complexity: The cognitive implications of biculturalism. Journal of Cross-Cultural Psychology, 40(1), 105–139.

Yudkin, D., Hawkins, S., & Dixon, T. (2019). The Perception Gap: How false impressions are pulling Americans apart. More in Common.

Share

Subscribe now

Leave a comment