AI is not arriving in a world capable of receiving it. Instead, it is coming into a world that has spent the last two decades systematically dismantling the very foundations that would allow us to deploy such a powerful technology responsibly. Institutional authority, shared truth, and information integrity have been battered. We are building the most sophisticated cognitive technology in human history atop the most fractured trust landscape in modern memory.
If you wanted to design the worst possible moment for AI to emerge, you might create a moment like this one.
A decade ago, we worried about living in a “post-truth” world. Oxford Dictionaries selected “post-truth” as its Word of the Year in 2016, defining it as circumstances in which objective facts are less influential in shaping public opinion than appeals to emotion and personal belief. Post-truth describes a world where facts still exist but are often overshadowed by emotional narratives.
What we are experiencing now goes deeper. I call this the “post-trust era:” a world where the very mechanisms that allow us to establish trust have broken down. It is not just that emotions sometimes trump facts. It is that we have lost shared foundations for deciding what counts as a fact in the first place.
Look at what has happened to our information ecosystem over the past two decades. We moved from a world in which truth was curated by editors, printed by presses, and distributed by a small number of recognizable intermediaries, to a world in which anyone can publish, amplify, distort, or fabricate. Data exploded, and social media fragmented audiences. Algorithms began optimizing for clicks, engagement, and sensationalism. The incentives shifted from accuracy to attention: virality over veracity, outrage over nuance, speed over scrutiny.
The numbers tell their own story. Nearly 8 in 10 people get directed to their news primarily through algorithmic systems such as social feeds, search engines, and aggregators, effectively outsourcing their information diet to systems they neither understand nor fully trust. In the United States, trust in the federal government has fallen from over 70% in the late 1950s to under 20% today. Confidence in mass media has dropped: fewer than one-third of Americans express even a “fair amount” of trust. Religious institutions, financial systems, healthcare organizations, and even science itself have seen trust erode.
Post-truth describes a distortion in how people weigh facts and feelings. Post-trust describes a deeper fracture: the loss of shared procedures for determining what is real.
When the first photographs appeared in the 1830s, people marveled at their fidelity. For nearly two centuries, seeing was believing. A picture was not perfect, but it was evidence.
In early 2023, an AI-generated image of Pope Francis wearing a stylish white puffer jacket spread across social media platforms. It was compelling because it sat in a narrow band between familiar and unexpected. The photo’s style, posture, and lighting borrowed credibility from decades of real photography. Many accepted it as real before they had time to question it.
Within two years, AI-generated images, voices, and video had become so pervasive and so convincing that the question was no longer “can you spot the fake?” but “can you trust anything you see?”
The ability to fabricate convincing “evidence” at scale has arrived just as our information ecosystem is most fragmented. Research from Duke and NYU shows that exposing people to opposing political views on social media can increase polarization rather than reduce it. A landmark study in Science found that false news spreads significantly faster and more widely than trustworthy news online. Neuroimaging research suggests that when people encounter information challenging their political beliefs, brain regions associated with negative emotion light up, while confirmatory information activates reward pathways. We are not only divided about facts. We are wired to process confirming and disconfirming information through different cognitive channels altogether.
That is the substrate into which AI arrives: a world where information is abundant, but trust is scarce, where consensus reality has splintered, and where the mechanisms for establishing shared truth have broken down.
Into this fractured substrate arrives something genuinely new. Not just another tool, but the first technology in history whose core competence is cognition.
Past technologies extended our capabilities while remaining subordinate to human thought. The printing press amplified the words we chose to print. Electricity transformed the energy we decided to harness. Combustion engines replaced the muscle power we used, and computers executed calculations we programmed. The internet distributed the messages we wrote.
AI breaks that pattern. It no longer merely executes tasks. It generates arguments, synthesizes information, makes recommendations, and produces creative work that feels, at a human level, like someone else’s thoughts. A single model can write an essay, summarize a legal transcript, translate a medical note, draft code, plan a marketing campaign, and compose music.
At the heart of most of these systems is an architecture that learns by predicting what comes next, a word, pixel, or token, in a sequence. These models do not know “truth” in the way humans understand it. They discover patterns, frequencies, and correlations in whatever data they are given.
And here is where the timing becomes critical: the data they are given is generated in a post-trust world.
AI is trained on our news stories, our social media feeds, our digitized books, our biased archives, our polarized debates, our conspiracy theories, our scientific breakthroughs, and our misinformation. It absorbs not only our knowledge but our distortions. It learns not only our facts but also our conflicts about those facts.
The result is a strange mirror. AI reflects us to ourselves, refracted through statistical patterns and scaled by computation. It gives us answers we cannot easily audit, based on training data we cannot thoroughly inspect, optimized for objectives we do not always control.
A new kind of cognitive infrastructure is emerging at the precise moment our social infrastructure for trust is under strain.
Excerpted from The Trust Code with permission from Tiffany Xingyu Wang.
