What Now
- Rich Washburn
- 2 hours ago
- 9 min read
Sam Altman went on the Relentless podcast and said something that should have stopped the conversation cold.
"We are now in the singularity. This is the moment."
Not someday. Not in a decade. Not when we achieve some agreed-upon benchmark. Now. This is the moment.
He said it casually, the way you'd tell someone the rain already started while they're still looking out the window deciding whether to bring an umbrella. And then he described what that actually feels like from inside the machine room, using a metaphor that was better than it had any right to be. The genie.
Altman's framing had four points. The first three are the kind of thing you expect from an AI CEO talking to a podcast host. We are close to creating a genie that can grant any wish. We are going to try to make sure our first wishes broadly benefit humanity. The space of what you can wish for is incredibly big and creative, and it will be exciting to figure that out with real human values and preferences.
Fine. Standard Altman. Optimistic, careful, a little rehearsed.
Then he said the fourth thing, and that is the one that matters.
"You start making these wishes, the computer grants them, and then you're like, I didn't think that was gonna work. What now?"
He compared it to mathematics. "People spent like a hundred years trying to disprove some Jacobian this thing. And then some guy asked Claude. And it was like, I don't know, within a few days or something like this."
The mangled line — because Altman was speaking off the cuff, not reading a paper — was this: people spent almost a hundred years trying to disprove the Jacobian conjecture. And then someone asked Claude. That someone was Levent Alpöge, a number theorist at Anthropic. Working with an experimental Claude model code-named "Fable," Alpöge produced a concrete counterexample to the Jacobian conjecture — a problem that had been open since 1939, when the German mathematician Ott-Heinrich Keller first posed it.
The counterexample was 216 characters long. It was independently verified in less than 24 hours. Terence Tao checked it. The result held. Eighty-seven years of mathematical effort. Reduced to a string of symbols shorter than a tweet.
I need to stop here and be precise about what the Jacobian conjecture actually is, because the details matter. The conjecture says, roughly, that if you have a polynomial map — a function built from addition and multiplication — and its Jacobian determinant (a measure of how the map distorts space) is a nonzero constant everywhere, then the map must be invertible. You can undo it. There's a polynomial map that takes you back.
It sounds technical. It is technical. But the intuition is simple: if a polynomial transformation doesn't collapse space anywhere — it preserves dimension everywhere — then you should be able to reverse it. It should have an inverse that is also a polynomial.
For 87 years, mathematicians assumed this was true. They proved special cases. They developed entire research programs around it. They built careers on the surrounding architecture. The problem appeared in Stuart Pin's "A Survey of the Jacobian Conjecture" in 1982. It showed up in conference proceedings, in graduate seminars, in the background assumptions of other work. It was part of the furniture. And then a number theorist and a language model found a counterexample. A polynomial map in three dimensions whose Jacobian determinant is a nonzero constant — specifically, negative two — everywhere it is evaluated, yet the map is not invertible. The conjecture is false.
The construction was not found by a human alone, nor by a machine alone. Alpöge designed the search pipeline. Claude Fable generated candidate constructions. Alpöge filtered, directed, and verified. The exact division of labor still deserves careful formal documentation, as several mathematicians have noted. But the result itself was checked and confirmed within a day. This is the part Altman was talking about. Not the genie metaphor. The "what now."
Here is what I have been thinking about since I heard the clip.
For most of history, human ambition has been disciplined by friction. You could imagine anything, but execution was expensive, slow, and uncertain. It required expertise, institutions, capital, labor, permission, and years of persistence. Those constraints did not merely prevent us from doing things. They helped us decide which things were worth doing. The friction was a filter. Hard problems served as a kind of triage — the difficulty itself told you something about which questions mattered, which careers made sense, which institutions deserved funding. The genie removes that filter. And suddenly the central problem changes. It is no longer "how do I accomplish this?" It becomes "what should I ask for?" And immediately after that: "now that it exists, what do I do with it?"
That is a much stranger problem than unemployment. Altman said he thinks there will be lots of great jobs in the future, lots of intellectual fulfillment, and that most jobs will adapt more than it seems like they should. I think he may be right about that. But then he said something else that was more honest: mathematics is an important example of something that may not go that way. In fact, probably won't go that way. He's right, and the reason he's right is specific.
Mathematics is not merely employment for mathematicians. It is a culture built around the pursuit of problems that may resist generations of human effort. The difficulty is structural. It determines status, specialization, mentorship, institutional funding, and even the emotional rhythm of a mathematical life. The career arc of a research mathematician — graduate school, postdoc, tenure, deep work on hard problems over decades — is organized around the assumption that the problems will hold.
Suppose machines begin resolving those problems routinely. Not merely checking proofs, but generating the decisive counterexample or the key conceptual move. Mathematics does not disappear. But the human role moves very fast. From proving to selecting problems. From searching to specifying search spaces. From constructing answers to validating and interpreting them. And eventually from asking existing questions to inventing questions worth asking. That sounds like adaptation. It may not be a gentle adaptation. A field can remain intellectually important while most of its former workflow becomes obsolete. The mathematician who spent thirty years developing intuition for polynomial maps wakes up to find that a model can test ten thousand candidates in an afternoon. The intuition is not worthless. But it is no longer the bottleneck.
Altman is quietly contradicting the standard comforting line that AI will simply give everyone better tools. A genie is not a better hammer. A genie collapses the distance between intention and consequence. That creates at least three new human bottlenecks.
The first is taste. When creation becomes cheap, judgment becomes scarce. Most wishes will be trivial, derivative, incoherent, or actively harmful. Knowing what deserves to exist becomes more valuable than knowing how to construct it. The skill that mattered in the era of scarcity was execution. The skill that matters in the era of abundance is discernment.
The second is specification. A genie grants the wish you actually communicate, not necessarily the future you vaguely imagined. The better AI becomes at execution, the more dangerous ambiguity becomes. "Make education better" is not a plan. It is an invitation for hidden assumptions to become infrastructure. The gap between what you meant and what you said has always existed. It used to be absorbed by the humans in the loop — colleagues, contractors, employees who could interpret intent and fill in the gaps. When the execution layer is a machine, that gap becomes a cliff. Precision of intent becomes the primary operational skill.
The third is meaning after success. Human beings are remarkably good at building purpose around unsolved problems. We organize careers around them. We organize institutions around them. We organize our sense of self around them. I am the person working on this problem. I am the field that studies this question. We are the community that has been pursuing this for three generations.
We are less practiced at dealing with a world in which the problem disappears in three days.
That "what now?" feeling may become one of the defining psychological conditions of the next era. Not unemployment. Not obsolescence. Something stranger. The discovery that your entire identity was organized around the impossibility of a thing, and then the thing happened.
The Jacobian example is almost perfect for this.
Humanity carried a question for 87 years. Mathematicians built careers, departments, and research programs around it. The conjecture was named in the 1970s, decades after Keller posed it, because by then it had become significant enough to deserve a name. Generations of mathematicians learned it in graduate school. Some tried to prove it. Some tried to disprove it. Most simply assumed it was true and built on top of it.
Then a person and a machine reduced the central answer to a construction that can be checked by hand. The counterexample is 216 characters. You can write it on an index card. The achievement is extraordinary. It also reveals something unsettling. A large portion of civilization has been psychologically calibrated around the assumption that difficult things will remain difficult for a long time. Not because we are lazy or uncreative. Because the difficulty itself was the structure. It told us what to do with our lives.
The genie doesn't merely grant wishes. It destroys the old timetable between desire, labor, and reality.
I wrote about this in May, in a piece called "The Cave by the Ridge." The argument then was about Plato's allegory — prisoners in a cave watching shadows on a wall, mistaking the projections for reality. I connected it to an earlier result where an OpenAI model had disproved a conjecture that Paul Erdős posed in 1946. Eighty years. Gone in an afternoon.
The connection to the Jacobian result is direct. The same pattern. A problem that defined a field for generations. A machine that dissolves it while you're getting coffee. The cave is still there. The shadows are still moving. But someone has turned around and seen the fire, and the fire is a language model running in a data center, and it is solving the things that made the cave feel like home.
Altman's singularity declaration makes more sense in this context. A year earlier he had written an essay called "The Gentle Singularity." The thesis was that the singularity would not arrive as a cinematic event. No flash in the sky. No dramatic moment where everything changes. Instead, it would arrive gradually, unevenly, across different fields at different speeds, while ordinary life remained deceptively recognizable. The grocery store stays open. People still commute. Most institutions behave normally. Meanwhile, inside technical fields, tasks that carried an assumed difficulty of months, years, or generations begin collapsing into prompts and machine-verifiable outputs. That is what "we are now in the singularity" means. It does not mean the world has visibly transformed. It means the accelerating process people historically called the singularity is already underway, and has been for some time, and the evidence is accumulating in technical fields that most people don't follow.
The Jacobian conjecture is evidence. The Erdős conjecture was evidence. The fact that Altman can sit on a podcast and describe the early singularity as an lived experience rather than a prediction is evidence. And the part that Altman got right, that most coverage missed, is the 99 percent problem. "And yet 99% of people outside our bubble don't get what's happening." Not because 99 percent of people are stupid. Because the gentle singularity doesn't announce itself. It just keeps dissolving hard problems while the rest of the world goes to work.
Altman said something else on that podcast that I want to pull out. He said he thinks there will be lots of great jobs in the future, lots of intellectual fulfillment. And then he said mathematics probably won't go that way.
That is a more significant admission than it sounds. Mathematics is the canonical intellectual discipline. If you had to pick one field that seemed safe from automation — one field where the argument "AI will help humans think better, not replace them" seemed most defensible — you would pick mathematics. The complexity, the creativity, the need for deep intuition built over years of study. Mathematics was supposed to be the fortress. If Altman is saying the fortress may not hold, he is saying something about the range of fields that will. If mathematics is a field where the human role moves rapidly from proving to selecting, from searching to specifying, from constructing to validating — then that pattern is not limited to mathematics. It is the pattern. Every intellectual field that has been organized around the difficulty of the work is going to experience the same shift, on some timeline. The question is not whether the shift happens. The question is what humans do on the other side of it.
Once the timetable collapses, the most important human skill may no longer be intelligence in the traditional sense. Not problem-solving. Not analysis. Not even creativity, if creativity means generating novel combinations of existing ideas — because machines can do that too, faster and at greater scale. The skill that matters may be the ability to decide what is worth wanting. And to remain human after getting it.
That sounds abstract. It is not. It is the most practical question the AI era has produced. When the genie can grant almost any wish, the genie stops being the bottleneck. The bottleneck is you. The bottleneck is whether you can articulate what you actually want with enough precision and enough wisdom that the result is something you would choose to live inside. Most people have never been forced to answer that question, because for most of history the friction did it for them. You couldn't wish for everything, so you didn't have to decide what you really wanted. The difficulty was the filter. The scarcity was the structure.
Now the filter is gone. The structure is gone. And what's left is a person and a machine and a question that has suddenly become urgent: What do you wish for? And after the wish is granted: What now?
Rich Washburn is a technologist and strategist working at the intersection of AI, infrastructure, and capital. He is Managing Partner and Chief AI Officer at Eliakim Capital.
Comments