In a long-form reflection, Alberto Romero writes from the fifth-floor balcony of his hotel in Lucerne about the accelerating, uneven advance of artificial intelligence and what it feels like to live through that change. The lake, swans and the Alps represent a reassuring continuity, but his phone and timeline bring a different tempo: weekly breakthroughs and warnings that make it feel as if years are now compressed into weeks.
The scene and an inner alarm
Romero contrasts the tranquil, centuries-old landscape — the Lake of the Four Forest Cantons, waterfowl, and the immobile Alpine peaks — with the frantic online stream: calls for government action on AI safety, lab employees warning about unknown consequences, and one high-performing model after another. He describes the sensation of being caught with one hand in a "black hole" while the other remains in Lucerne, watching summer slip away.
Speed and uneven distribution
He observes that people label this acceleration in many ways — the Singularity, exponential growth, a Cambrian explosion — but stresses that the name is less important than the lived experience of being carried along without agency. Crucially, Romero emphasizes that AI’s impact is highly uneven: access, model quality, capital, compute, technical knowledge, usage frequency, and willingness to delegate all vary. That means the same technology can feel like nothing to a casual user and like an existential rupture to a domain expert.
Quoting Scott Aaronson, Romero notes that sometime in the past year we passed a significant line: progress is now reported in front-page news rather than only debated in arguments. Aaronson’s important qualification is that the singularity has already started but is “wildly unevenly distributed.”
Will AI remain spiky or become uniform?
Romero frames AI today as a "spiky star": high peaks of capability interspersed with low floors. He asks whether that shape will persist or whether AI could evolve into a more circular, uniformly powerful force that flattens differences across domains — a development that would undermine the protective walls many people rely on (taste, embodied skills, non-verifiable human experiences).
He admits his own refuge — personal, sensory, flesh-based experiences that seem to resist automation — may be brittle. If that wall is not durable, the consequences for identity and meaningful work would be profound.
Voices of acceptance, resilience and resistance
Romero surveys other thinkers. Scott Aaronson leans toward acceptance tempered by the reality of uneven distribution. Erik Hoel argues for active resistance: he worries that with real superintelligence, discoveries, decisions and corporations would be run by banks of machines, and he endorsed a ban on superintelligence in Washington to prevent being relegated to subservience.
Others defend remaining human roles: writers and critics such as Jasmine Sun, Adam Mastroianni and Sam Kriss argue that AI lacks lived experience, taste and an authorial voice. Romero sympathizes with that skepticism but also recognizes that these arguments cannot prove certainty about the future.
Inside the industry: Shengyu Liu’s grief
Romero highlights Shengyu Liu, presented as a kernel engineer at DeepSeek, who has worked on making models cheaper, faster and more efficient. Liu’s account is personal: he expects to change his work when kernels can be written better by AI. He may still earn a living, but he may lose the work he loves. That loss is not only professional — it’s the breaking apart of a life where what you do, what you are paid for and what you love coincide.
Romero points out that this grief reflects a broader anxiety: progress freed many from manual toil, but at scale, automating skilled, meaningful tasks raises the question of whether there is a point where further progress diminishes human flourishing.
Love, taste and the meaning of work
Throughout the essay, Romero returns to love, small human pleasures and the idea that having interested people matters more than objectively interesting things. He quotes Annie Dillard and recalls earlier aphorisms: "there are no interesting things, only interested people" and "we will always have the kisses." These remarks are an argument for preserving what gives life meaning even as technologies change the conditions of labor and creativity.
Yet he refuses to romanticize: not everyone can become an "aristocrat" of leisure once machines take over. For many, the rhythm of work is integral to identity, and its loss is real.
Where to draw the line?
Romero rejects absolutist positions — neither "stop development today" nor unfettered progress is a self-evident answer. The central question he raises is whether there exists a boundary where further development does more harm than good, and whether humanity can recognize that boundary. He does not claim to have the answer but insists the dilemma must be addressed rather than ignored.
A closing scene: rain on the lake
The piece ends where it began, in Lucerne. A sudden downpour and fog swallow the lake, the valley and even the outlines of the Alps. For Romero that moment is emblematic: the trip’s last day and the metaphorical clouding of certainties. The intellectual debate has turned into personal mourning and reflection — gratitude for finitude, empathy for those losing their crafts, and an unsettled fear about what the future will bring.
Romero’s essay stitches together personal scenes, industry testimony and philosophical questions about agency, identity and the distributional nature of AI’s effects. It does not resolve the tensions it raises, but it insists they be felt and discussed rather than shrugged off.
