Seeing Things That Aren't Quite There

People constantly act on judgments they can't spell out: he gives off a creepy vibe, this email looks like phishing, something about this deal is off. In an essay for Aeon, the philosopher Marc A Moffett argues that acting on such reactions is often fully rational. His case rests on the claim that intuition and perception share a feature, though it isn't the feature philosophers usually point to.

The neglected kind of intuition

Moffett argues that these exotic cases are continuous with ordinary imagining. Suppose you are deciding whether to tell a friend his husband has been unfaithful. You picture his hurt, and that pain seems like a reason to stay quiet. You don't infer this and you don't literally see it. It presents itself. The trolley and barn verdicts arrive the same way: unbidden, before any deliberation, and we trust them.

Why perception is the model

The puzzle is where that trust comes from. Perception has a known causal route: light reaches the eyes. Intuition has no organ and no obvious object it makes contact with.

Two problems with "it's how it feels"

Presentationality as a posture

Moffett's answer is that presentationality is a cognitive posture, a way of taking in a content, not a feeling. He places it between two familiar attitudes. Belief is assertive: it commits you to the truth of something. Imagination is non-assertive: it entertains a possibility without treating it as real. A seeming sits in the middle. It presents something as actual without committing you to it, so it offers a route to belief that is direct but can be overturned.

Seemings are also happenings, not actions. You don't decide to have an intuition any more than you decide to see a tree. Moffett argues that this passivity is what lets a seeming present its content as actual. Once something is presented that way, believing it is a mind's default response. In his words, belief-formation is "less like drawing a conclusion and more like yielding to a disclosure."

On this view the absent-intuition challenge goes away. No special intellectual glow is needed. Someone can feel nothing distinctive and still grasp a content as actual.

Intuition inside perception

Moffett then argues that intuition does work inside ordinary perception. Drawing on P F Strawson, he notes that we see a child reaching for a toy or a car coming too fast, not patches of colour. His example: you and a trained herpetologist, a snake expert, see the same snake, and your sensory experiences may be identical. To you it is "a snake". To her it is Agkistrodon contortrix, the copperhead. The difference lies in the concepts at work, not in what you see. He says this is why expertise can change experience without changing how things look.

He also answers a sceptic who says this is just skill at applying concepts. Applying a concept, such as judging that Gollum betrays Frodo in The Lord of the Rings, is a judgment, and judgments need a basis. The intuition that the scene is a betrayal comes before the judgment and makes it reasonable. "Intuition does not compete with concept application. It enables it."

What remains open

Moffett's conclusion is moderate. Vibes are neither "vindicated wholesale" nor "dismissed as mere noise". They are fallible, but they are the same kind of state that underlies all our basic contact with the world. A thought experiment is that same capacity, tightly controlled and aimed at one concept.

The essay leaves several practical questions unanswered. It doesn't say how to tell a trustworthy hunch from a biased one, beyond noting that seemings can be defeated. It doesn't engage the experimental work showing that thought-experiment verdicts vary across people and framings. And defining presentationality as a "posture" names the property more than it explains how the mind produces it.

Writing for Big Think, Namir Khaliq treats that post as a symptom. A growing set of writing argues that AI could create a "permanent underclass": everyone who fails to entrench themselves before superintelligence pushes the value of human labour toward zero. Khaliq went to the Bay Area to find out why young, bright people had turned so pessimistic, and what they were doing about it.

A meme with an academic past

The phrase is older than the internet. Economist Gunnar Myrdal popularized "underclass" in the 1960s. In 1979 the Carnegie Council on Policy Studies in Higher Education warned of "a permanent underclass," a self-perpetuating culture of poverty that would form if young people weren't readied for jobs. Online, the phrase was linked to AI by at least 2024. In summer 2025 it took off as a sardonic joke.

Clark says he never meant any of it. "I just don't think a lot of technologists in the Bay Area understand how basic economics work," he told Khaliq. In June he broke character to say the underclass "doesn't exist and it will all be fine." Khaliq notes that the meme still pays Clark's bills, and asks whether the rest of the workforce is in on the joke.

Two camps, one mansion

Stanford's status panic

At Stanford, sophomore Saheb Gulati has become the person friends text when they worry about the underclass. He thinks the fear hits hard there for a reason. Your "entire self-concept," he says, rests on joining the American elite, so being told you'll be knocked off the top "is very taxing on your psyche." Khaliq links this to Yale law professor Daniel Markovits's argument that meritocracy traps elites in an endless struggle to avoid "elimination."

The people already falling

Khaliq's former roommate quit Big Tech to start an AI company. AI has cut build times "to two days, if not hours," and "software engineering feels done. If you stay in your job, you're kind of a sitting duck." He agreed with a viral post by venture capitalist Deedy Das. Das wrote that about 10,000 people had passed $20 million in wealth over five years, while everyone earning under $500,000 felt they could never catch up.

"Paul" is a pseudonymous Midwestern engineer with more than 20 years in the field. He was laid off from a major tech company and has had no paycheck for over 18 months. He thinks AI was more scapegoat than cause: blaming "forces beyond our control" plays better in the press than admitting to cost-cutting. Either way, he says, "I'm probably the underclass now."

Joan Williams, a law professor at UC Law San Francisco who studies economically precarious people, sees a familiar pattern. AI may do to white-collar workers what automation and outsourcing did to blue-collar workers. The software workforce skews young and male, which she calls a social tinderbox: "There's nothing so dangerous as a man without a future."

Bostrom's other worry

Philosopher Nick Bostrom made the AI-doom argument academically respectable with Superintelligence (2014). His newer book, Deep Utopia, asks what people would do if machines handled everything. He expects a painful transition: graduates may see their own wages fall while some blue-collar wages rise, and parents may feel all that homework "will be for naught." He thinks the deeper obstacle is cultural. People have tied self-worth to being useful. Quoting Keynes ("we have been trained too long to strive and not to enjoy"), he proposes schools that teach conversation, friendship, art and spirituality, which he calls being "great at leisure."

Where it leaves things

Khaliq ends with Gulati, who argues for solidarity over escape. In Gulati's view, post-AGI questions about health, education and democracy have "essentially zero people working on them," so the work is to rebuild institutions rather than to get yourself out personally. The piece gives no data on how big AI's actual effect on jobs is. Its evidence is the anxiety itself, gathered in one unusually anxious region.

The Public Domain Review, a non-profit that publishes out-of-copyright art and texts with short essays, has collected nine Indian paintings made between about 1700 and 1900. All of them show Krishna, the Hindu god, and Rādhā, his cowherd lover, wearing each other's clothes. The unsigned essay that comes with them argues that the images express one particular idea of love: a wish to remove every difference between two people. It also argues that the wish can never be fully met.

A grammar of fusion

Gurevitch's point is that the English hides something. Sanskrit has a dual number, a grammatical form for exactly two things, alongside singular and plural. Kālidāsa doesn't write "Pārvatī and Shiva" as two words joined by "and". He writes pārvatīparameśvarau, one compound word in the dual. Pārvatī's name loses its own case ending, and the whole word takes the gender of its last part. Inside the grammar, the goddess and the god become a single unit. The comparison to signifier and signified works the same way: a word's sound and its meaning can't be pulled apart once you understand the word.

The essay takes this as an example of a theme found all through Indian literature and art: closing the gap between two bodies. Sometimes the gap is huge. In Kālidāsa's Meghadūta ("Cloud Messenger"), a separated lover sends a message to his wife across the whole subcontinent. Sometimes the gap is just the plain, unbearable fact that two people are two and not one.

What the paintings show

The paintings handle the swap in several ways. In some, the lovers simply look at each other in their borrowed clothes, tenderly or amused by their own game. In others they also swap roles. Rādhā sings to Krishna or leads him as they walk. He puts his crown on her head and lets her fix his hair.

Empathy, not complementarity

The cycle of separation

The essay argues that the pull between them is matched by stronger forces that separate them. It says scenes of Rādhā heartbroken are probably as common as scenes of bliss. Krishna has left to join his wives, or he is flirting with other gopīs. When she is angry, the roles flip again. The god is humbled and sent away by a mortal woman. He has to use go-betweens to plead for him and has to kneel to stroke her feet in apology.

These quarrels follow a fixed pattern. Reconciliation is certain, and so is the next offence. The cycle "has recurred already many times over." The essay ends with Gurevitch: "Even when we are perfectly close to each other, a yearning, longing, desire, to separate, recombine, be together in different ways continues to exist." Read this way, the clothes swap is one move in an endless game of coming together and parting, not a final merger.

What the piece leaves open

The argument

Public talk about dark matter usually asks whether it exists at all. In a Big Think essay, astrophysicist Ethan Siegel argues that this is the wrong question. He calls the people who doubt dark matter "an extraordinarily slim minority" and says the issue has been settled for decades. The question working researchers now ask is which kind of dark matter fills the universe. Siegel's main point is that astronomy can answer this without a laboratory ever detecting a dark matter particle.

Why researchers consider existence settled

Siegel lists several independent lines of evidence, and they agree with one another:

Siegel says no single alternative explains all of these unless it adds something that behaves exactly like dark matter. The particle itself has never been detected, and he notes there is no guarantee it ever can be.

Hot, warm or cold

The next question is what kind of dark matter it is. Astrophysicists sort the candidates by how fast the particles were moving early in cosmic history compared with their mass:

• Hot: light particles moving close to the speed of light for much of cosmic history. Fast particles stream out of small clumps before they can collapse. The result is shallow gravitational wells on small scales and very little small structure.
• Warm: fast at first, then slowing down at intermediate times. This gives medium-depth wells.
• Cold: particles that were either heavy or born slow, so they could clump very early. This gives the deepest wells, even on small scales.

As a rough analogy (not Siegel's), think of trying to pile sand versus marbles shot from a cannon. Only slow material settles into small heaps. Maps of the cosmic web, the network of filaments and clusters that galaxies form, already rule out hot dark matter and most warm versions. Siegel says at least 98% of dark matter must be cold or warm.

The "spherical cow" problem

Four ways to see the clumps

2. Intracluster light. In galaxy clusters, stars get stripped from galaxies and drift between them. Over long periods these loose stars settle into the cluster's overall gravity, so their faint glow traces both the main halo and its sub-halos. Hubble only first captured this light in 2018, in the cluster Abell S1063. Deeper, sharper images of more distant clusters could show how substructure changes over cosmic time.

4. Stellar streams. When the Milky Way's tidal forces tear apart a globular cluster (a dense ball of old stars) or a small satellite galaxy, its stars spread out into a long stream. A dark sub-halo passing through a stream disturbs it. Several streams at different distances from the galactic center can map how dark matter is spread out and how much of it sits in clumps. Siegel says that in 2026 astronomers found the first globular-cluster stream outside the Milky Way, which suggests the method could work in other galaxies too.

What's left open

This is an essay arguing a position, not a report of new results. Siegel doesn't say which model the evidence currently favors beyond "cold or mostly cold." He doesn't explain how the measurements turn into a particle mass. He also mentions only briefly that simulations leave out the effects of gas and stars. That gap matters: ordinary matter reshapes halo centers, so it can be hard to tell whether a mismatch with observations points to new physics or to a missing ingredient in the simulation. His larger claim, that the existence debate no longer interests researchers, is his own judgment. The article gives no figures on how widely that view is shared.

One revolution, not six

Palmer starts by asking why Silicon Valley keeps reaching for the Renaissance and the age of empires. Her first answer is that people in a crisis search the past for alternatives. French revolutionaries dressed as imagined Romans, and the Renaissance followed the Black Death. Her second answer is that these are the periods we link with disruptive technology: the printing press and the Industrial Revolution. Her third is about fiction. Empire is easy to write because power sits with a few dramatic characters. "Just try writing fiction about the European Union," she says.

Then comes the line the piece is named after. Silicon Valley people ask her whether we have really had "six Gutenberg revolutions in a row, 10 years apart." She says no. The personal computer, the internet, the cell phone, social media and generative AI are all stages of one change, the computer. In the same way, Gutenberg's press kept producing new effects for decades, through news networks, then pamphlets, then newspapers. Her complaint is that commentators analyse the last five years when they should be looking at the last twenty.

The guy behind the refrigerator

Liu says job losses to AI were set up long before AI arrived, through what he calls the "gradual mechanization of humans." His example is customer service. Forty years ago it was a respected profession. It was then broken down into scripted tasks, and once a job is a script, swapping in a language model is easy. Palmer offers an analogy for this. You fill a new refrigerator with something strange, such as synthetic meat, and the fridge struggles. Everyone blames the new food. Meanwhile a man behind the fridge has been hitting it with a sledgehammer for weeks. In her analogy the sledgehammer stands for decades of weakening unions, skills and employee rights.

Collier objects that this future depends on automating a lot of hard work. How do humans keep real power? Liu answers that you can no more give power to AI than to your refrigerator. The risk is handing power to the few people who control AI, and the fix is the old one: spread power out and stop states from monopolizing knowledge. Palmer argues that the wheel, horses and washing machines all cut drudgery without weakening anyone. What weakens people is institutions and policy.

Two categories of people

Palmer splits the fear of idleness into two claims. A: many people would stop working if they could. B: idle people are dangerous. She traces B to Machiavelli's Discourses, where he agrees with the Roman historian Livy that a state should be founded somewhere hard, to keep ordinary people busy. She also points to a double standard. The same readers enjoy novels about idle Regency aristocrats. That suggests a belief that some people can be trusted with leisure and others cannot. She argues that humanity keeps reinventing aristocracy, and that much of what Silicon Valley calls meritocracy is a new version of it. Liu's caveat is that some people get their happiness from ranking above others. How many such people there are, and whether they could break the system, is unclear.

Keep the useful 10%

Liu calls most office uses of AI, emails and PowerPoint, "absolutely useless." He sees the real value in science: protein folding, theorem proving, searching chemical possibilities. Collier pushes back from both sides. Software engineers say AI has transformed their work, while many biomedical scientists say the benefits are oversold. Palmer borrows Cory Doctorow's terms. In a "centaur" setup the human is in charge and the machine helps. In a "reverse centaur" the human serves the machine, for example by correcting its output. People in the first setup tend to like AI and people in the second tend to hate it. Her small-scale example is Bowling Green, Kentucky. The city used a language model to sort about 8,000 survey responses its tiny staff could never have read. The top issues were funding for an English-as-a-second-language program and one dangerous intersection, and neither was on any politician's radar. Her middle-ground model comes from radium. We once put it in face cream and cough syrup, then dropped the roughly 90% of uses that hurt us and kept power, medicine and space systems.

Despair is how we lose