The ascending friction thesis — that every technological abstraction relocates difficulty to a higher cognitive floor rather than eliminating it — is the load-bearing wall of the optimistic case for AI. Vetlesen's framework subjects the thesis to a test it has not yet passed: is the higher-level difficulty genuinely constitutive, producing the sedimentary understanding that lower-level difficulty produced, or is it simulated difficulty — the surface behavior of depth without the geological foundation that gives depth its weight? The distinction separates the two cases is phenomenological: genuine difficulty produces discomfort, the informationally rich state of encountering what resists, while simulated difficulty produces the appearance of engagement without the discomfort that signals constitutive encounter.
The laparoscopic surgery example supports the ascending friction thesis because the ascending friction there is structural — built into the physics of the new medium, inescapable, demanding of anyone who enters the field. The surgeon cannot avoid the higher-level difficulty even if she wants to. The instruments require new skills; the screen demands new perceptions. The difficulty ascended because the medium forced it to ascend.
The AI-assisted knowledge worker inhabits a different environment. The medium does not force difficulty to ascend. It offers a choice — a choice the builder may not recognize as a choice, because the frictionless path and the friction-rich path produce outputs that, from the outside, look identical. The lawyer who drafts a brief by hand and the lawyer who reviews AI output both produce competent briefs. One's judgment is constituted by the encounter with the material; the other's is a review function, applied to output but not formed by the process.
The diagnostic test is phenomenological, not behavioral. Genuine difficulty produces the discomfort of not-knowing, the sustained engagement with what resists, the experience of sitting with a problem that will not resolve itself. Simulated difficulty produces the appearance of engagement — scanning output, making corrections, approving results — without the discomfort that signals constitutive encounter. From the outside the two are indistinguishable. From the inside, only the practitioner who has experienced both can tell.
Merleau-Ponty's phenomenology of embodied cognition provides the deeper stakes. Understanding is constituted by the body's engagement with a world that resists, that has texture, that demands specific attention. When AI writes the code or drafts the brief, the body is removed from the epistemic loop. The engagement is real, but it is not embodied in Merleau-Ponty's sense — which means the understanding that develops may be structurally different from the understanding that develops through direct engagement with the material.
The concept is this volume's specific contribution, extending Vetlesen's framework to interrogate The Orange Pill's ascending friction thesis. The term 'ascending anesthesia' names the possibility that the optimistic case conceals — the shadow thesis that must be ruled out empirically before the ascending friction claim can be accepted as general.
Structural vs. elective friction. Laparoscopic surgery forces ascending difficulty; AI-assisted work offers it as a choice. Elective frictions tend to be declined.
Constitutive vs. review judgment. Judgment that develops through encounter with material differs from judgment applied to already-generated output. The first grows with use; the second is consumed.
The phenomenological test. Discomfort — the informationally rich state of sitting with what resists — is the diagnostic signature of constitutive friction. Engagement without discomfort may be ascending anesthesia.
The invisibility of the difference. Flow and compulsion, embodied judgment and review judgment, ascending friction and ascending anesthesia are indistinguishable from the outside. The distinction requires first-person phenomenological attention the culture is discouraging.
Proponents of the ascending friction thesis argue that the phenomenological test is too demanding: we cannot run it at civilization scale, and by the time we could run it, the damage would be done. Vetlesen's response is that the damage is already observable in specific populations (the Berkeley study, the deskilling evidence) and that the burden of proof lies with the party claiming ascending friction, not with the party insisting the question be asked.